0 study of various mac protocols for wsn

Review

A meticulous study of various medium access control protocolsfor wireless sensor networks

Ratnadip Adhikari

School of Computer and Systems sciences, Jawaharlal Nehru University, New Delhi 110067, India

a r t i c l e i n f o

Article history:

Received 25 February 2013

Received in revised form

22 November 2013

Accepted 27 January 2014

Keywords:

Data communication

Wireless sensor networks

MAC protocols

Distributed nodes

Energy waste

Collision avoidance

Overhearing

a b s t r a c t

During the last decade, Wireless Sensor Networks (WSNs) have evolved as an incredibly useful

technology in the area of signal processing and data communication. They have found prolific

applications in a wide range of domains which include cell phone monitoring, robotic exploration,

disaster management, intrusion detection and medical systems. Medium Access Control (MAC) protocols

constitute an important set of regulations which enables the successful and smooth operation of the

WSN. A fundamental design goal of all MAC protocols is to prevent energy wastes from various possible

sources during data communications. Till now, a wide variety of MAC protocols with different objectives

have been accumulated in sensor network literature. A thorough study of these protocols is very

important both from the perspectives of understanding the current research trends and determining

scopes for further innovative works in this domain. This paper meticulously discusses about the

associated issues and difficulties which are faced in designing efficient MAC protocols for WSNs. Several

popular MAC protocols are described here with their inherent merits and demerits. In order to provide

an up-to-date survey, various MAC protocols which have been developed relatively recently are

discussed, together with the traditional benchmark ones. Finally, this paper concludes with outlining a

number of innovative ideas and future research directions in this domain.

& 2014 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2. Design and implementation issues of MAC protocols for WSN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1. Primary reasons of energy waste . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.2. Properties of a good MAC protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3. Various developed MAC protocols for WSN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3.1. MACA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3.2. MACAW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3.3. IEEE 802.11. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3.4. Power aware multi-access signaling (PAMAS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3.5. Sensor MAC (S-MAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3.5.1. An empirical demonstration of energy saving vs. increased latency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3.5.2. Advantages of S-MAC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3.5.3. Disadvantages of S-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3.6. Timeout MAC (T-MAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3.6.1. Clustering and synchronization in T-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3.6.2. RTS operation in T-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3.6.3. Determining the threshold TA in T-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.6.4. One solution of the early sleeping problem in T-MAC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.6.5. Advantages of T-MAC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.6.6. Disadvantages of T-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.7. Dynamic sensor MAC (DS-MAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jnca

Journal of Network and Computer Applications

http://dx.doi.org/10.1016/j.jnca.2014.01.011

1084-8045 & 2014 Elsevier Ltd. All rights reserved.

E-mail address: [email protected]

Please cite this article as: Adhikari R. A meticulous study of various medium access control protocols for wireless sensor networks.Journal of Network and Computer Applications (2014), http://dx.doi.org/10.1016/j.jnca.2014.01.011i

Journal of Network and Computer Applications ∎ (∎∎∎∎) ∎∎∎–∎∎∎

www.sciencedirect.com/science/journal/10848045

www.elsevier.com/locate/jnca




mailto:[email protected]





3.8. Eyes MAC (EMACs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.9. WiseMAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.9.1. Advantages of WiseMAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.9.2. Disadvantages of WiseMAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.10. Sift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.10.1. Advantages of sift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.10.2. Disadvantages of sift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.11. Optimized MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.12. Traffic adaptive medium access protocol (TRAMA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.12.1. Advantages of TRAMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.12.2. Disadvantages of TRAMA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.13. Self-organizing MAC (SMACs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.14. Energy aware TDMA based MAC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.15. Berkeley media access control (B-MAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.16. Data gathering MAC (D-MAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.16.1. Advantages of D-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.16.2. Disadvantages of D-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.17. Lightweight medium access protocol (LMAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.18. Pattern MAC (PMAC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.19. Zebra MAC (Z-MAC). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.20. X-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.21. Funneling-MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4. Discussions and future research directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

1. Introduction

An overwhelming raise in the demand for collecting and

utilizing information about the surroundings has been observed

throughout the last decade. A breakthrough in this domain is the

concept of Wireless Sensor Networks (WSNs) which can process

and disseminate knowledge in notably fast speed (Calí et al.,

2007). WSNs are beneficial over many other traditional networks

in terms of cost, size and efficiency. Over the years, these networks

have been used in diversified fields such as cell phone monitoring,

robotic exploration, intrusion detection, disaster management,

climate control, and temperature pressure monitoring (Yadav

et al., 2009; Akyildiz et al., 2002). A typical WSN consists of a

large number of sensor nodes which are distributed in the

environment to collectively constitute a multi-hop wireless net-

work. Each sensor node is composed of an embedded processor,

low power radio and limited memory unit. These nodes are

operated through batteries and are organized to perform a

common task (Ye et al., 2002). Due to low power capacities of

the sensor nodes, a WSN has limited coverage and range for

communication as compared to other mobile devices. Thus, such a

network must contain large number of interconnecting nodes for

successful practical applications.

Sensor networks have different issues and challenges depending

on the situations they are applied for. One crucial challenge faced is

energy consumption. It is often very difficult to change or replace the

exhausted batteries of the constituent nodes in a sensor network

which is an obvious obstacle in maximizing the network lifetime. In

order to reduce the energy consumption, a major objective of a

sensor network is to minimize the associated communication while

achieving the desired network operation (Yadav et al., 2009;

Demirkol et al., 2006). Extensive research works have been carried

out on the design of low power electronic devices to reduce energy

consumption in sensor networks. However, due to hardware limita-

tions and manufacturing costs it has been observed that substantial

energy consumption can be more economically achieved through

designing energy efficient communication protocols.

It is an incredibly challenging task to create a wireless sensor

network that implements energy efficient medium access rules

among the heavily populated low capacity sensor nodes (Demirkol

et al., 2006). Medium Access Control (MAC) is an important

technique for the successful and smooth operation of a sensor

network. MAC specifies a set of rigorous rules which enables the

associated WSN to perform the desired network operations in an

energy efficient manner. Designing highly effective power saving

MAC protocols is a major way to considerably prolong the sensor

network's lifetime. Also, one of the fundamental objectives of a

typical MAC protocol is to prevent interfering nodes from colliding

while communication (Yadav et al., 2009; Ye et al., 2002). In this

regard, a number of attributes must have to be considered while

designing a good MAC protocol for a WSN (Ye et al., 2002). The

first is the energy efficiency. The nodes in a WSN are assumed to be

dead when they are out of battery and as such the proposed MAC

protocol must be effective enough to reduce the potential energy

wastes during data transmissions. Next important attributes are the

scalability and adaptability to changes. The designed MAC protocol

should efficiently as well as rapidly handle the changes in network

topology, size and node density for a successful adaptation. Another

important attribute is the fairness. In traditional WSNs, each user

desires equal opportunity and time to access the medium and so

per-hop MAC level fairness is important. However, in sensor net-

works, all nodes cooperate for a single common task and normally

there is only one application; at a certain time, a node may have

considerably more data to send than some other nodes. Thus,

fairness is not so important as long as application-level performance

is not degraded. Other attributes, e.g. latency, throughput and

bandwidth utilization may be useful but are secondary in WSNs.

Designing MAC protocols for WSNs is an active research area

having important contributions from numerous researchers. Dur-

ing the last decade, several MAC protocols have been developed

for wireless voice and data communication networks which can be

broadly classified into two major categories: contention based and

schedule based protocols (Ye et al., 2002). Looking at the increasing

number of MAC protocols which have been accumulated in the

sensor network literature over the past few years, a thorough and

systematic study of them is very important from the perspective of

future innovative works in this domain. This paper is destined to

meticulously discuss about a wide variety of MAC protocols for

R. Adhikari / Journal of Network and Computer Applications ∎ (∎∎∎∎) ∎∎∎–∎∎∎2





WSNs, emphasizing their relative strengths and weaknesses.

Various associated issues and challenges, such as collision avoid-

ance, idle listening, and overhearing are lucidly discussed. In order

to provide an up-to-date survey, several recently proposed MAC

protocols are described together with the well-known benchmark

ones. Finally, the paper provides a comparison of the studied MAC

protocols and outlines a number of innovative ideas and potential

future research directions in this domain.

The rest of this paper is organized as follows. Section 2

discusses the major challenges which are faced in a MAC protocol

design and points out the characteristics as well as different

evaluation metrics with regards to a potentially good MAC proto-

col for WSN. Section 3 describes a wide variety of MAC protocols

together with their inherent merits and demerits. In this section,

altogether 21 MAC protocols are studied, including both the well-

known traditional ones and those which have been developed

fairly recently. Then, Section 4 concisely compares the studied

MAC protocols and outlines a number of open issues and future

research directions. Finally, the paper is concluded in Section 5.

2. Design and implementation issues of MAC

protocols for WSN

The prime objective of a MAC protocol is to substantially

increase the lifetime of the WSN through reducing the potential

energy wastes during communication. To achieve this goal, a

proposed MAC protocol must address and attempt to resolve all

sources of energy waste in a sensor network operation. In this

section, we discuss the major reasons for energy consumption and

summarize the properties which a good MAC protocol for WSN

must satisfy.

2.1. Primary reasons of energy waste

In a sensor network consisting of a dense population of battery

powered nodes, it is very difficult to maintain the required energy

supplies to all nodes. Hence, increasing the lifetime of the WSN is

of course a crucial challenge. The energy consumption in a WSN

occurs in three domains (Calí et al., 2007): sensing, data processing

and communication. The major sources and reasons of energy

waste are discussed below (Calí et al., 2007; Yadav et al., 2009;

Demirkol et al., 2006; Van Dam and Langendoen, 2003):

� Collision: When a receiver node receives more than one packet

at the same time, these packets are then called collided packets.

All packets which cause the collision have to be discarded and

the subsequent follow-on retransmissions considerably

increase energy consumption. The occurrence of collision

increases latency as well.� Overhearing: It refers to the phenomenon where a sensor node

picks up the packets which are intended for some other nodes.

This results in wastage of network energy.� Protocol overhead: Control packets in a WSN do not contain

useful information and so their transmissions consume unne-

cessary energy.� Idle listening: A major source of energy waste in a WSN is idle

listening, i.e. listening to receive possible traffic from an idle

channel which is actually sending nothing. This happens in

most WSNs as a node goes to an idle state if nothing is sensed.� Over emitting: Energy waste by over emitting occurs through

transmission of a message when the intended destination node

is not yet ready.� Adaptation: It refers to reconfiguring a WSN when nodes join

or leave the arrangement and is also a major source of

energy waste.

2.2. Properties of a good MAC protocol

A well-designed MAC protocol must have to satisfy certain

important properties. The primary among them is to effectively

prevent the energy wastes from all the aforementioned possible

sources (Yadav et al., 2009; Demirkol et al., 2006). A MAC protocol

should also possess a number of additional characteristics, such as

(i) energy efficiency, (ii) scalability and adaptability to changes in

the network properties, (iii) fairness, (iv) latency and (v) throughput

and bandwidth utilization. Moreover, there should be robust

measures in order to evaluate as well as compare the perfor-

mances of various developed MAC protocols. In this regard, the

following evaluation metrics are widely used by WSN researchers

(Yadav et al., 2009).

� Bitwise energy consumption: It is defined as the ratio of the total

energy consumed by a sensor node and the total number of bits

transmitted. Obviously, for two MAC protocols, the one having

a lesser value of this metric is more energy efficient than

the other.� Average packet latency: It measures the average time taken by

the dispatched packets to reach the destination nodes. The

lesser the value of this metric, the better is the performance of

the designed MAC protocol.� Average delivery ratio: It is defined as the average ratio of the

number of packets received to the number of packets sent for

all sensor nodes.� Throughput: The throughput of a network is a count of the total

number of packets delivered to the destination node per unit

of time.

These four metrics are very useful in assessing the performance

of a designed MAC protocol. The numerical values of these metrics

measure the competence of a MAC protocol in terms of energy

efficiency, latency and throughput.

3. Various developed MAC protocols for WSN

For a WSN, designing a good MAC protocol that satisfies the

characteristics as discussed in Section 2 is indeed a challenging

task. There are several obstacles, encountered while routing in

such networks. The major among them is the architecture of the

sensor network itself. A typical WSN is composed of a large

number of low-powered sensor nodes which are constrained in

supplies of energy, bandwidth, storage and processing ability

(Akkaya and Younis, 2005). The constituent sensor nodes are in

general stationary having non-rechargeable and irreplaceable

limited capacity batteries. These factors make them prone to

severe energy deficiency during data communication. The deploy-

ment of sensor nodes in a typical end-to-end IP-enabled WSN

architecture is shown in Fig. 1. Second, the classical IP-based

protocols are inapplicable in sensor networks as it is practically

impossible to create a global addressing scheme for such a

network. Third, in almost all sensor networks, sensed data from

multiple sources are communicated to one particular sink. This is

contrary to other familiar communication networks, such as

wireless ad hoc and wired networks. Fourth, in order to improve

energy and bandwidth utilization, an efficient routing protocol

must have a well-defined mechanism to exploit data redundancy.

Such redundancies are common in a sensor network as different

sensors may generate the same data within a neighborhood.

Considering the mentioned challenges, any routing protocol for a

WSN in a nutshell should effectively address these six issues

(Akkaya and Younis, 2005; Tilak et al., 2002): (i) network dynamics,

R. Adhikari / Journal of Network and Computer Applications ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 3





(ii) deployment of sensor nodes, (iii) node capabilities, (iv) energy

consumptions, (v) data delivery schemes and (vi) data aggregation or

fusion.

Although MAC protocols for WSNs is a relatively new domain of

research, but a number of important contributions in this topic

have already been made which can be found in literature. Till date,

there exist several MAC protocols, designed with different per-

spectives (Yadav et al., 2009; Ye et al., 2002; Demirkol et al., 2006;

Van Hoesel and Havinga, 2004). A systematic classified study of

these developed MAC protocols is very essential for further

innovations in this area. Present MAC protocols for WSNs can be

broadly categorized into two major classes: contention based and

schedule based protocols (Yadav et al., 2009; Ye et al., 2002). The

contention based protocols are mostly based on the Carrier Sense

Multiple Access (CSMA) technique and are popular due to their

simplicity and robustness. These protocols can easily adjust to the

network topology changes and also have relaxed time synchroni-

zation requirements. However, these are affected by large amount

of energy wastes through collisions, overhearing and idle listening.

A recent study has found that the energy consumption using

the contention based protocol IEEE 802.11 is very high due to

MAC protocols

for WSN

Contention based Schedule based

Technology

CSMA IEEE 802.11 TDMA Bluetooth

Example Technology Example

Easily adaptable to

topology changes

Relaxed time

synchronizations

Much energy

consuming

Highly energy

efficient

Strict time

synchronizations

Hard to adapt to

topology changes

Merits Demerit Merit Demerits

Fig. 2. Classification of MAC protocols for wireless sensor network.

A B C

Fig. 3. Three communicating nodes in a network configuration.

Fig. 1. Deployment of sensor nodes in a typical end-to-end IP-enabled wireless sensor network.






significant idle listening (Stemm and Katz, 1997). On the other

hand, the scheduled based protocols are based on the Time

Dependent Multiple Access (TDMA) technology and are considerably

more energy efficient than their contention based counterparts.

The scheduled based protocols require strict time synchroniza-

tions and avoid collisions, overhearing and idle listening among

the sensor nodes through scheduling transmit and listen periods.

However, in TDMA protocols, e.g. Bluetooth (Haartsen, 2000) and

LEACH (Heinzelman et al., 2000), the nodes form various clusters

and managing the communications among such clusters is a quite

difficult task. A TDMA protocol cannot easily modify its frame

length and time slot arrangements according to the changes in the

number of clusters or the number of nodes within the clusters.

Hence, its adaptability to the changes in network topology is not as

efficient as that of a contention based protocol. The classification

of MAC protocols for WSNs is depicted in Fig. 2.

In this section, we discuss about various important MAC

protocols which are proposed in literature for WSNs. We start

with the well-known traditional protocols and gradually advance

towards the more recent ones. The essential features of the

proposed MAC protocols, together with their advantages and

disadvantages are presented in a lucid manner. For sake of their

great importance, some benchmark MAC protocols are described

with more details.

3.1. MACA

This protocol was proposed by Karn (Karn, 1990) as an alter-

native to the traditional CSMA technology for sensor networks.

Unlike the carrier sense methodology, the exchange between

Request-to-send (RTS) and Clear-to-send (CTS) packets in MACA

enables collision avoidance at the receiver side and not at the

sender. This protocol is briefly explained with the help of the

network configuration with three nodes A, B and C, as shown

in Fig. 3.

The fixed size signaling packets: RTS and CTS, both of which

contain the proposed length of the data transmission are used in

communication through MACA. Station A willing to transmit to

station B sends an RTS to station B to which station B replies

immediately with a CTS if it is ready to hear and not deferring at

that moment. Station A sends the intended data to station B

immediately after receiving the CTS. The stations overhearing the

RTS and CTS packets defer all transmissions up to a predefined

duration of time. Station A times out ultimately if it does not hear

the expected CTS from station B, assuming that a collision has

occurred and then schedule the packet for retransmission. The

retransmission time in MACA is selected through the binary

exponential back-off (BEB) algorithm (Bharghavan et al., 1994).

MACA ensures that a station hearing an RTS will wait enough

so that the transmitting station can receive the returning CTS. On

the other hand, any station hearing the CTS will avoid colliding

with the returning data transmission. The stations which hear an

RTS but not a CTS can perform data transmissions without any

harm after the CTS has been sent. This is because such stations are

not prone to collision as they are out of range of the receiver.

Let us now demonstrate the working of this protocol in a

practical scenario. In the network configuration of Fig. 3, station B

can hear both A and C, but stations A and C cannot hear each other.

The hidden terminal scenario occurs when C attempts to transmit

while A is transmitting to B. Similarly, an exposed terminal scenario

occurs when B is transmitting to A while C attempts to transmit.

Employing MACA in the hidden terminal scenario, station C would

not hear the RTS from station A but would hear the CTS from

station B and so would defer from transmitting during A's

transmission. On the other hand, in the exposed terminal scenario,

station C would hear the RTS from station B but not the CTS from

station A and thus would be free to transmit during B's data

transmission.

MACA is simple to understand and implement. It often success-

fully avoids collisions during data transmission. It also effectively

handles the hidden and exposed terminal scenarios. But, the main

disadvantages of this protocol are the high energy consumption

due to idle listening, lack of synchronization of RTS–CTS transmis-

sions, and lack of adequate fairness level in many one-cell

communication configurations (Bharghavan et al., 1994).

3.2. MACAW

MACAW is a slotted MAC protocol, proposed by Bharghavan

et al. (1994) as a modification to the MACA protocol. In this

protocol, collision avoidance is achieved through introducing Data

Sending (DS) and Acknowledgment (ACK) packets together with the

usual RTS/CTS packets of MACA. MACAW uses an RTS–CTS–DS–

DATA–ACK message exchange scheme and includes a back-off

algorithm that is considerably different from the one used by the

MACA protocol.

MACAW protocol was mainly introduced to extend the func-

tionalities of the MACA protocol and it does not use the carrier

sensing methodology but is based on a different approach.

MACAW requires a sensor node to send a DS packet just before

sending the actual data for making the neighboring nodes aware

of the fact that the RTS–CTS exchange was successful. In order to

explain the working principle of MACAW, we consider a simple

network configuration with five sensor nodes: A, B, C, D and E, as

shown in Fig. 4.

We assume that the node B wants to transmit data to the node

C in the network configuration of Fig. 4. Then, a possible successful

data transmission using MACAW involves the following sequence

of steps:

1. B sends an RTS to C which is also heard by A.

2. C replies with the CTS which is also heard by D.

3. B now sends the DS to A to inform that the RTS–CTS exchange

was successful.

4. B then starts sending the data packet to C.

5. Now, let E attempts to transmit data to D and so sends an RTS

to D. D hears the RTS but will not reply due to the ongoing

transmission to C.

6. After the transmission was successful, C sends an ACK to B.

7. Now, D is allowed to transmit the data and so sends a Request

for RTS (RRTS) to E so that E becomes aware of the successful

transmission and the idle channel.

It should be noted that MACAW adopts a non-persistent

mechanism. This means that if the transmission contains more

than one packet, A has to wait for a random time after every

successful data transfer and then again has to compete with the

adjacent nodes for gaining access to the transmission medium.

MACAW is widely popular due to its simple yet rigorous packet

transmission mechanism and many MAC protocols for WSN are

Fig. 4. Communication pattern of the MACAW protocol.






based on it. It provides considerably more effective collision

avoidance as compared to MACA. It is also easily adaptable to

changes in network topology and supports varieties of commu-

nication patterns. Furthermore, MACAW is also known to provide

robust solution to the hidden terminal problem (Ye et al., 2002).

However, like MACA, energy consumption in MACAW is often high

due to idle listening of the constituent sensor nodes. Another

problem faced by MACAW is the lack of synchronizing information

in some situations. Although, the RRTS packet is introduced as a

remedial measure, but still it cannot solve all such contention

based problems (Bharghavan et al., 1994).

3.3. IEEE 802.11

The IEEE 802.11 is a familiar contention based MAC protocol

that is primarily developed on the methodology of MACAW. It uses

carrier sensing and randomized back-offs to avoid collisions of the

data packets. The contention mechanism in IEEE 802.11 is same as

that of MACAW. It uses the usual RTS and CTS packets to manage

transmission. The sensor node that first sends the RTS packet gains

access to the medium and the receiver node will respond with a

CTS packet. Once the data transmission is started, the nodes do not

follow their sleep (i.e. idle listening) schedules until the transmis-

sion is completed. The IEEE 802.11 protocol reduces the idle

listening by periodically entering into the sleep state through the

Power Save Mode (PSM) (Yadav et al., 2009). This protocol also

supports fragmentation of the communication packets. The RTS

and CTS packets reserve the medium only for the first data

fragment and the first ACK. The first fragment and ACK then

reserves the medium for the second fragment and ACK and so on.

Thus, each neighboring node receiving a fragment or an ACK

knows that there is one more fragment to be sent and so it keeps

listening until all the fragments are sent. But, this mechanism is

amenable to a lot of energy waste through overhearing by all

neighboring nodes.

The IEEE 802.11 is widely used in wireless ad hoc and sensor

networks due to its straightforward methodology, simple time

synchronization, and collision avoidance mechanism. For single-

hop networks, it reduces a lot of energy waste from idle listening

by using the PSM technique. This protocol is also found to provide

simple and robust solution to the problem of hidden terminal (Ye

et al., 2002). However, it has a very high energy consumption rate

due to overhearing. The fragmentation mechanism lets a node to

unnecessarily wait even if no further fragment or ACK has to be

received. Also, all the neighboring nodes overhear the transmis-

sion between two nodes in the network. All these processes waste

a lot of precious energy.

3.4. Power aware multi-access signaling (PAMAS)

PAMAS (Singh and Raghavendra, 1998) is one of the earliest

contention based MAC protocols, specifically designed for wireless

ad hoc networks with energy efficiency being its primary goal.

This protocol later influenced the design of many other advanced

MAC protocols, such as the Sensor MAC (S-MAC) (Ye et al., 2002).

This protocol deliberately turns off the sensor nodes which are not

receiving or sending any data in order to conserve energy. A sensor

node sets its radio to sleep during transmissions of other nodes.

PAMAS uses two separate channels for RTS/CTS control and

data packets. A node with a packet to transmit sends an RTS over

the control channel and waits for the CTS from the receiving node.

The transmitting node enters into a back-off state if no CTS arrives

for a predefined duration and if a CTS arrives within the specu-

lated time, the transmitting node sends the data over the data

channel. The receiving node transmits a busy tone over the control

channel to let the other nodes in the network know about its busy

status. The use of two radios at every sensor node in the different

frequency bands in PAMAS leads to considerable increase in the

sensor cost, size and design complexity. Also, excessive switching

between sleep and wakeup states results a lot of energy

consumption.

Fig. 5. The S-MAC mechanism: (a) periodic listen/sleep and (b) inter-cluster

synchronization.

Fig. 6. The communication mechanism in S-MAC.

Fig. 7. Energy saving vs. average sleep delay for S-MAC protocol.






3.5. Sensor MAC (S-MAC)

The S-MAC (Ye et al., 2002) is a contention based MAC protocol,

specifically designed for wireless sensor networks as a significant

enhancement of the earlier IEEE 802.11 protocol. S-MAC is con-

ceptually inspired by the PAMAS protocol, but unlike PAMAS, it

does not require an additional channel for communication. It uses

three novel techniques, viz. periodic listen–sleep, virtual clustering

and message passing in order to reduce energy consumption and

support self-configuration. The sensor nodes periodically wake up,

communicate data and again return to sleep. During the sleep

period, the sensor nodes turn off their radios and after the sleep

period, they wake up to listen whether communication is

addressed to them. If the awaked nodes find that there is nothing

communicated towards them and they have their own data to

transmit, then they can initiate communication themselves. The

sleep and listen periods are locally synchronized among the nodes.

However, this synchronization is not very strict and the nodes can

use their sleep period for communication if needed. Neighboring

nodes form virtual clusters to set up a common sleep schedule.

Two neighboring nodes, residing in two different clusters wake up

at listen periods of both the clusters. The periodic listen/sleep and

inter-cluster synchronization are illustrated in Fig. 5(a) and

(b) respectively. As can be seen from Fig. 5(b), the nodes A and B

belong to two neighboring clusters and so they synchronize with

nodes C and D, respectively. If A wants to communicate with B, it

just waits until B is still listening. Similarly, if multiple neighboring

nodes want to communicate with one particular node, they must

have to contend for the medium when the node is listening.

Schedule exchange and period synchronization in S-MAC are

accomplished by sending a short SYNC packet which includes the

address of the sender and its next time to sleep. The nodes which

receive this SYNC packet, immediately update their timers and a

node goes to sleep when its timer expires. The period for every

node to send a SYNC packet is called the synchronization period.

The listen interval of a node is divided into two parts so that it can

receive both synchronization and data packets. Each part is further

divided into many time slots for adequately performing carrier

sensing. An example of communication through S-MAC is shown

in Fig. 6 which was given by Ye et al. (2002). This figure depicts the

timing relationship between a receiver and several senders, where

CS stands for carrier sense.

Collision avoidance which is basic to any MAC protocol is

performed through a carrier sense scheme in S-MAC. Both virtual

and physical carrier sense together with RTS/CTS exchange is

performed in S-MAC (Ye et al., 2002). For virtual carrier sensing, a

duration field called Network Allocation Vector (NAV) is introduced in

each transmitted packet. The NAV indicates how long the remaining

transmission will be. When a sensor node wants to communicate, it

first looks at the NAV and if the value of the NAV is non-zero, then it

is understood that the medium is busy. Physical carrier sensing is

performed at the physical layer by listening to the communication

channel for possible transmissions.

S-MAC introduces an effective policy to substantially reduce

overhearing. Each node calculates and maintains its NAV value to

indicate the activity of the neighboring nodes. When a node

receives a packet which is actually destined to other nodes, it

immediately updates its NAV by the duration field in the packet. In

this manner, the node realizes the time for which it has to remain

silent. A non-zero NAV value means that active transmissions are

in process in the neighborhood. Also, the NAV value decrements

every time when the NAV timer fires. So, in order to avoid

overhearing, a node should sleep if its NAV is not zero and can

wake up again when its NAV becomes zero.

A very important feature of S-MAC is the concept of message

passing where long messages are divided into small frames and

sent in a burst. The disadvantages of transmitting a long message

as a single packet are the high cost of re-transmitting if only a few

bits have been corrupted in the first transmission. Through

fragmentation of the long message into many independent small

packets, considerable energy saving can be achieved by minimiz-

ing communication overhead at the expense of unfairness in

medium access (Ye et al., 2002; Demirkol et al., 2006).

Another salient feature of S-MAC is the adaptive listening.

Periodic sleeping may result in high latency, especially for multi-

hop routing algorithms. Adaptive listening improves the latency

caused by periodic sleeping (also known as sleep delay) and in turn

improves the overall latency (Demirkol et al., 2006). This scheme

requires that an overhearing node wakes up for a short duration at

the end of the transmission, so that if it is the next-hope node,

then its neighbor could pass the data immediately. The duration

field of the RTS/CTS packets is used by the sensor nodes to know

the end of the active transmissions.

3.5.1. An empirical demonstration of energy saving vs.

increased latency

S-MAC provides reasonably large amount of energy saving,

however with the expense of increased latency and as such an

analysis of the tradeoff between these two factors is useful. A

packet moving through a typical multi-hop network experiences

the following delays at each hop: carrier sense delay, back-off delay,

transmission delay, propagation delay, processing delay and queuing

delay. These six delays are common to any contention based MAC

protocol. S-MAC introduces the additional sleep delay which is the

delay caused by periodic listening. The effect of this delay was

analyzed by Ye et al. (2002) which we present here.

Let us consider that a packet arrives at the sender with equal

probability of time for listen and sleep. Also, let a frame refers to

one complete cycle of listen and sleep. Then, the average sleep

delay Ds on the transmitting node is

Ds ¼ T f rame=2 ð1Þ

where

T f rame ¼ T listenþTsleep ð2Þ

In the above equations, Tlisten, Tsleep and Tframe denote the listen,

sleep and frame time, respectively. Now, comparing with the

protocols which do not have the periodic sleep mechanism, the

relative energy saving Es in S-MAC can be given as

Es ¼Tsleep

T f rame

¼ 1�T listen

T f rame

ð3Þ

The last term in Eq. (3) is the duty cycle of the respective node.

It is desirable to have the listen time as short as possible, so that

Fig. 8. Communication patterns of the MAC protocols: (a) S-MAC and (b) T-MAC.






for a certain duty cycle, the average sleep delay is short. The

percentage of energy saving Es vs. sleep delay Ds on each node for

the listen time of 300 ms and 200 ms is depicted in Fig. 7.

From Fig. 7, we can see that even if the sleep time for a node is

zero which indicates no sleeping, there is still a delay. This

happens because of the fact that contention only starts at the

beginning of each listen interval.

3.5.2. Advantages of S-MAC

S-MAC is conceptually simple to understand and implement,

yet provides very good energy conserving compared to various

other contention based protocols, especially the IEEE 802.11. In

S-MAC, energy waste is substantially reduced as well as time

synchronization overhead is prevented to a large extent through

sleep schedules. This protocol also does not require very strict time

synchronization and the sensor nodes can even use their sleep

periods for communicating. Another salient feature of S-MAC is

that it is able to make tradeoffs between energy saving and latency

as per the network traffic situation.

3.5.3. Disadvantages of S-MAC

Although straightforward but S-MAC allows a low duty cycle

unless the active time is significantly smaller than the wakeup

period (Anastasi et al., 2009). Another drawback of the S-MAC

protocol is that when two neighboring clusters follow entirely

different schedules, then a lot of energy is consumed from idle

listening and overhearing. Furthermore, the sleep and listen

periods in S-MAC are predefined constants which results in high

latency, low throughput and less efficiency under variable traffic

load (Demirkol et al., 2006).

3.6. Timeout MAC (T-MAC)

As stated earlier, one major drawback of S-MAC is the constant

listen and sleep periods which lead to low duty cycle and high

latency. The static listen/sleep period also considerably degrades

the performance of S-MAC under variable network traffic load. In

S-MAC, a sensor node remains awake for the whole awake period

even if there is no active communication. The T-MAC protocol (Van

Dam and Langendoen, 2003) is designed to overcome these

drawbacks of S-MAC and to enhance its overall performance. T-

MAC increases the duty cycle through shortening the awake

period of the idle channel. It lets the sensor nodes to turn off

their radios if no activity is detected for at least a timeout value

(Anastasi et al., 2009). A node listens to the channel for the

predefined timeout period and if no data is received during this

period, it returns to sleep mode. But, if some data is received, then

the node remains awake until the timeout or awake period,

whichever is shorter. The communication mechanism of T-MAC

is briefly explained below.

T-MAC reduces idle listening and overhearing by transmitting

all messages in bursts of variable lengths and allowing nodes to

sleep between bursts. All messages are properly queued and nodes

communicate with each other using a RTS–CTS–DATA–ACK

scheme. In order to maintain an optimal active time under variable

load, the concept of timeout period is additionally introduced. The

diagrammatic comparison between S-MAC and T-MAC commu-

nication pattern is shown in Fig. 8(a) and (b).

3.6.1. Clustering and synchronization in T-MAC

Like S-MAC, T-MAC adopts virtual clustering for frame syn-

chronization (Van Dam and Langendoen, 2003). A node begins

waiting and listening, once it becomes active. If it hears nothing

for certain duration of time, it chooses a frame schedule and

transmits a SYNC packet which contains the time until the next

frame starts. On the other hand, if the node hears a SYNC packet

from other nodes during startup, it follows the schedule in that

packet and transmits its own SYNC accordingly. Nodes must listen

for a complete SYNC frame sporadically to be aware of the

existence of different schedules. This allows new and mobile

nodes to adapt to an existing group.

If a node having a schedule of its own hears a SYNC with a

different schedule, then it has to adopt both schedules and also

has to transmit its own SYNC to the other node, so that the other

node can know the presence of another schedule. In this manner,

through adopting both schedules, the node will have an activation

event at the start of both frames. Nodes must start a data

transmission only when their own active time starts. At that time,

both neighbors have the same schedule and also the neighbors

that have adopted the schedule as extra are awake.

The virtual clustering technique is easy to implement in T-MAC.

It requires communicating nodes to form clusters with the same

schedule without enforcing this schedule to other nodes in the

network. This mechanism provides efficient broadcast without

requiring the need to maintain information on individual

neighbors.

3.6.2. RTS operation in T-MAC

In T-MAC, each node transmits its queued messages in a burst

at the start of the frame and during this burst, the medium is

Fig. 9. A basic data exchange operation in the T-MAC protocol.

Fig. 10. The early sleeping problem with the T-MAC protocol.

Fig. 11. Depiction of a solution to the early sleeping problem in T-MAC.






saturated and messages are transmitted at maximum rate. A node

may have to fight fiercely every time for winning the medium

when it sends an RTS. An increasing contention interval is not

useful as the load is often very large and remains constant

throughout. Thus, RTS transmission in T-MAC starts by waiting

and listening for a random time within a fixed contention interval

which is intentionally tuned for the maximum load. The conten-

tion time is always used, even if no collision has occurred yet.

Further, the listen period ends when no activation event has

occurred for a predefined time threshold TA which is known as

the timeout period. Selection of TA must cope with the early

sleeping problem, described later. In Fig. 9, we present the

illustration of a basic data exchange in T-MAC.

In Fig. 9, it can be seen that node C overhears the CTS from

node B and will not disturb the communication between A and B.

The threshold TA must be long enough in order to allow C to hear

the start of the CTS.

3.6.3. Determining the threshold TA in T-MAC

T-MAC requires that a node should not go to sleep while its

neighbors are still communicating, since it may be the receiver of a

subsequent message. But, there is possibility that a node may fall

asleep earlier than the speculated time and this is known as the

early sleeping problem (Van Dam and Langendoen, 2003). The

broken synchronization among listen periods within the virtual

clusters is the main reason for this undesired problem (Demirkol

et al., 2006). An example of early sleeping is depicted in Fig. 10, in

which we can see that the node D goes to sleep before C can send

an RTS packet to it. In order to prevent this early sleeping problem,

the T-MAC threshold period TA must have to be determined

adequately.

Receiving the start of the RTS/CTS packet from a neighbor is

enough to trigger a renewed interval TA. But, a node may not hear

this because it lies outside the range and as such the threshold

period TA must have to be long enough so that a neighboring node

at least hear the start of the CTS packet if it did not hear the RTS

packet, in case. This observation gives us a lower limit on the

length of the interval TA:

TA4CþRþT ð4Þ

where C and R are respectively the lengths of the contention

interval and an RTS packet, whereas T is the turn-around time, i.e.

the short time between the end of the RTS packet and the

beginning of the CTS packet.

3.6.4. One solution of the early sleeping problem in T-MAC

The early sleeping problem often substantially degrades the

overall performance of the T-MAC protocol. The usual solution

which is used in T-MAC to guard against this problem is to

introduce a Future Request to Send (FRTS) packet together with

the usual RTS/CTS packets. The idea is to let another node know

that there is still a message for it but the sender itself is prohibited

from using the wireless sensor medium.

The scheme works as follows. If a node overhears a CTS packet

destined for another node, it may immediately send an FRTS

packet. The FRTS packet contains the length of the blocking data

communication which was in the CTS packet. A node must not

send an FRTS packet if it senses communication right after the CTS

or if it is prohibited from sending due to a prior RTS or CTS. A node

that receives an FRTS packet knows that it will be the future target

of an RTS packet and must be awake by that time. For the FRTS

solution to work, the threshold time TAmust be increased with the

length of the CTS packet. An illustration of the discussed solution

to the early sleeping problem is depicted in Fig. 11. In this figure,

we can see that the CTS overhearing node C immediately sent an

FRTS to the node D which keeps D awake.

3.6.5. Advantages of T-MAC

T-MAC resolves the static listen/sleep problem which is a major

source of energy waste in the S-MAC protocol. The performance of

S-MAC under variable traffic load is considerably improved in T-

MAC. The proposers (Van Dam and Langendoen, 2003) have

shown that T-MAC uses only one fifth of the energy used by S-

MAC for variable traffic loads. Moreover, the T-MAC is also based

on a loose time synchronization requirement.

3.6.6. Disadvantages of T-MAC

In spite of better results of T-MAC under variable traffic loads, it

leads to the early sleeping problem which is due to the broken

synchronization among listen periods in virtual clusters. Although,

a solution to this early sleeping problem is provided, as described

in Section 3.6.4, but it is not a robust one. The radio sensitivity

limits the range of overhearing and so the nodes lying outside the

range of ongoing communication go to sleep. This limits the data

forwarding process of T-MAC only to a few hops (Anastasi et al.,

2009). Furthermore, the reduction of energy waste for variable

traffic loads through adaptive duty cycling of T-MAC actually

comes at the undesired cost of reduced throughput and increased

latency.

3.7. Dynamic sensor MAC (DS-MAC)

DS-MAC (Lin et al., 2004) attempts to improve the performance

of S-MAC by decreasing the latency for delay-sensitive applica-

tions through adding a dynamic duty cycle feature. During com-

munication, all sensor nodes start with the same duty cycle and

within the SYNC period, all nodes share their one-hop latency

value among themselves. In this manner, DS-MAC doubles the

duty cycle in order to keep the schedules of the neighbors

unaffected (Demirkol et al., 2006). A node shortens its sleep time

and announces it through the SYNC packet whenever it notices a

high value of the average one-hop latency. The sender node which

receives this decrease in sleep period signal checks its queue traffic

for possible packets which are intended to that receiver node. If

there is one such packet, the sender node decides to double its

duty cycle when its battery level is higher than a specified

threshold value.

The duty cycle doubling in DS-MAC is shown in Fig. 12. It has

been observed that the latency of DS-MAC is considerably better

than that of S-MAC. It has been also observed that DS-MAC has a

better average power consumption per packet (Lin et al., 2004).

3.8. Eyes MAC (EMACs)

EMACs, developed by Van Hoesel et al. (2004) is a TDMA based

MAC protocol which consists of a fully distributed and self-

organizing scheme for communicating data in the wireless sensor

network. Each active node in EMACs periodically listens to the

channel and broadcasts a short control message. This control

message is used for medium access operations and also for gaining

useful information at reasonably low energy costs. The informa-

tion, contained in the control message is used to create a maximal

independent set of nodes from the network which in turn builds a

connected network of nodes. All nodes in this set are the active

nodes, whereas the remaining others are the passive nodes. The

active nodes communicate with each other in a collision-free

manner. The passive nodes on the other hand do not actively

participate in the communication and are specifically used for

energy saving through utilizing the framework of the connected

network. In this manner, EMACs reduces the protocol overhead for

passive nodes to a large extent and hence saves a lot of energy.

Although, the passive nodes may communicate with active nodes,






but this communication is not guaranteed to be collision-free.

Another salient feature of EMACs is its self-organizing nature, due

to which it does not depend on a base station or central manager.

Like TDMA, EMACs divides the communication time into time

slots which the participating nodes can use for data transmission

without requiring to contend for the medium. However, unlike

TDMA, the time slots are not managed by a central manager in

EMACs. As usual, a sensor node can assign only one time slot to

itself which it fully controls. After the specified frame length which

contains several time slots, a node again has a period of time

reserved for it. The nodes select their own time slot on the basis of

local information only. A time slot for an active node is further

divided into three sections: Communication Request (CR), Traffic

Control (TC) and the Data section. In the CR section, an active node

listens for the incoming requests from the passive nodes. Then, in

the TC section, it transmits a short control message which contains

various valuable control and synchronization information. Nodes

also hear their neighboring TCs for gaining knowledge of the

communication situation in the neighborhood. Finally, the data

section can be used for the actual transmission of data. An

example of EMACs time slot division is shown in Fig. 13.

EMACs is straightforward, based on a flexible framework and

greatly energy efficient. The simulations, performed by its devel-

opers (Van Hoesel et al., 2004) demonstrate that EMACs was able

to prolong the WSN lifetime 30–55% in a static topology, whereas

it prolonged the network lifetime with a factor 2.9–4.2 in a

dynamic topology as compared to S-MAC.

3.9. WiseMAC

The WiseMAC (Enz et al., 2004) is a MAC protocol developed for

WSN which is similar to the previously developed spatial TDMA

and CSMA with preamble sampling (El-Hoiydi, 2004). It uses Non-

persistent CSMA (np-CSMA) with preamble sampling in order to

decrease idle listening. In preamble sampling, all sensor nodes

have two communication channels, viz. data channel and control

channel which are respectively accessed through TDMA and CSMA

(Yadav et al., 2009). The receiving node is alerted by a preamble

which precedes each data packet. All nodes sample the medium

with a common period, but their relative schedule offsets are

independent. The preamble size is initially set equal to the sample

period. After waking up if a node finds the medium as busy, it

samples the medium and continues to listen until it receives a data

packet or the medium becomes idle again. However, there is a

possibility of energy waste through over emitting as the receiver

may not be ready at the end of the preamble. Over emitting is

further increased with the length of the preamble and the data

packet as no handshake is performed with the intended receiver.

Preamble sampling is prone to a lot of power consumption due

to the predefined fixed length of the preamble. As a remedial

measure, WiseMAC attempts to dynamically determine this

length. Also, unlike preamble sampling, WiseMAC uses only one

communication channel to reduce the problem of idle listening. In

the WiseMAC scheme, the sensor nodes learn and refresh the

sleep schedules of their neighbors during every data exchange as a

part of the acknowledgment message. All transmissions are

scheduled on the basis of the knowledge of the neighbor's sleep

schedules. A random wake up time is also proposed to decrease

the collision possibilities.

The choice of the wake up preamble length is also affected by

the potential clock drift between the source and the destination

nodes. A lower bound for the preamble length is calculated as the

minimum of the sampling period Tw of the destination node and

the potential clock drift with the destination. Considering this

lower bound, a preamble length, Tp is chosen randomly. The

concept of WiseMAC protocol, as discussed above is concisely

depicted in Fig. 14, due to Enz et al. (2004).

3.9.1. Advantages of WiseMAC

The dynamic determination of the length of the preamble

makes WiseMAC notably efficient under variable network traffic.

The scheme for handling clock drifts in this protocol lessens the

need for external time synchronization. It has been found through

simulations that WiseMAC achieves reasonably better perfor-

mance than one of the S-MAC variants (Enz et al. (2004)).

3.9.2. Disadvantages of WiseMAC

WiseMAC allows decentralized listen/sleep schedules due to

which the neighbors of a sensor node have different sleep and

wake up times. This often results in redundancy in data transmis-

sion for broadcast type of communication which in turn leads to

higher power consumption and latency (Demirkol et al., 2006).

Moreover, WiseMAC suffers from the hidden terminal problem

just like spatial TDMA and CSMA with preamble sampling and so,

the chances of collisions are dramatically increased.

3.10. Sift

Sift (Jamieson et al., 2006) is a randomized CSMA based MAC

protocol which is primarily proposed for event-driven wireless

sensor networks. In many CSMA protocols, multiple nodes in the

same neighborhood often sense a common event and as such it is

sufficient to consider the report of only that particular subset of

nodes which sense that event. However, traditional CSMA proto-

cols, e.g. IEEE 802.11 do not possess adequate mechanism to

handle this situation and as such provide degraded latency and

throughput. This is the key motivating factor behind the develop-

ment of Sift. Unlike traditional CSMA, Sift does not use a time

varying contention window from which a time slot is randomly

Fig. 12. Duty cycle doubling in DS-MAC.

CR TC DATA CR TC DATA

Time slot

. . .

Fig. 13. Time slot division in EMACs.Fig. 14. Concept of the WiseMAC protocol.






picked by a node. Rather, it keeps a fixed size of the contention

window and each slot within this window is transmitted through

a carefully chosen non-uniform probability distribution. This

mechanism of Sift significantly reduces the latency for the delivery

of event reports.

In a shared mediumwhere N nodes sense an event and contend

to transmit on the channel at the same time, the goal of Sift is to

minimize the time taken to send the first R (RrN) of these

messages without collisions. It should be noted that R¼N corre-

sponds to the classical throughput optimization problem. Thus, the

basic idea of Sift is that out of the total N reports, only the first R

are most crucial and so these have to be relayed with low latency.

A truncated, increasing geometric distribution is used as the back-

off probability distribution in Sift for picking a transmission slot

within the fixed size contention window. Due to its implementa-

tion simplicity, Sift is proposed to form a lower-layer building

block for various other future MAC protocols for WSNs (Jamieson

et al., 2006).

3.10.1. Advantages of sift

Sift achieves very low latency with a large network size in the

expense of energy consumption. But, slightly increased amount of

energy waste may be justified when latency is the major factor of

the concerned network. The authors have shown using simula-

tions that Sift can provide a 7-fold latency reduction as compared

to IEEE 802.11 for network size of up to 500 sensor nodes.

Moreover, Sift is a simple window based CSMA protocol which is

notably easy to implement. As such, it has been proposed that Sift

can be used as a MAC layer filter for various other sensor

information dissemination protocols (Jamieson et al., 2006).

3.10.2. Disadvantages of sift

Sift increases energy consumption through idle listening and

overhearing. Idle listening occurs because all slots have to be

listened before transmitting. On the other hand, overhearing is

increased because nodes must have to listen till the end of an

ongoing transmission in order to contend for the next transmis-

sion. Moreover, Sift requires system-wise time synchronization

and as such it is quite costly to implement for the protocols which

do not use time synchronization (Demirkol et al., 2006).

3.11. Optimized MAC

The optimized MAC protocol (Yadav et al., 2008) attempts to

reduce energy consumption from idle listening, control packet

overhead and overhearing through controlling node latency on the

basis of network traffic. The duty cycles of the nodes are changed

based on the traffic load in a direct proportionate manner. This

means that the duty cycle of a node is more if the traffic is more

and is less if the traffic is low. The pending queue of messages is

used for identifying the extent of network traffic. The control

packet overhead is minimized by reducing the number and size of

the transmission packets. For reducing their size, the source and

destination addresses are removed from data and control packets.

Optimized MAC improves the performance of S-MAC through

adaptive duty cycle and control packet overhead minimization.

S-MAC achieves significantly increased energy saving through

compromising sensor latency and as such, it may not be suitable

for delay sensitive applications. Optimized MAC, on the other hand

is good for applications where lower latency is a key goal apart

from energy efficiency. The obtained simulation results by the

authors (Yadav et al., 2008) show that optimized MAC is highly

energy efficient under a wide range of traffic loads and is also able

to adapt itself to improve sensor latency for congested network

traffic.

3.12. Traffic adaptive medium access protocol (TRAMA)

TRAMA (Rajendran et al., 2003) is a TDMA based MAC protocol,

developed for collision free energy efficient channel access in

WSNs. Reduction of energy waste in TRAMA is achieved by

ensuring that unicast and broadcast packets never collide and

allowing the sensor nodes to enter into a low-power idle state

whenever they are transmitting or receiving nothing. It employs a

traffic adaptive election scheme to select the receivers on the basis

of the schedules announced by the transmitters. Nodes exchange

their two-hop neighborhood information and transmission sche-

dules, indicating the intended receivers of their traffic. Then, the

nodes which should transmit and receive during each time slot are

selected in a proper order. This adaptive election scheme of

TRAMA ensures that all nodes within one-hop neighborhood of

the transmitter will receive data collision freely. It also successfully

eliminates the problem of hidden terminal. In a nutshell, TRAMA

consists of three components: Neighbor Protocol (NP), Schedule

Exchange Protocol (SEP) and Adaptive Election Algorithm (AEA).

A single, time-slotted channel is assumed in TRAMA for both

data and signal transmissions. Time is organized as random access

(known as signaling slot) and scheduled access (known as trans-

mission slot) periods. Signaling slots are used to establish two-hop

topology information, whereas transmission slots are used for

collision free data exchange and schedule propagation. The length

of a transmission slot is fixed with respect to channel bandwidth

and data size. As the signaling packets are normally smaller than

the data packets, so the transmission slots are set as a multiple of

the signaling slots for easy synchronization. A node announces its

intended slots and the intended recipients for these slots using a

schedule packet. The intended recipients are indicated in a sche-

dule packet by using a bitmap whose length is equal to the

number of neighbors of the corresponding node. A node broad-

casts that it renounces the allotted slots if it does not have

sufficient data to transmit. These vacant slots can be used by other

nodes for data transmission. The priority of a node on a particular

slot is calculated through a hash function. The time of the last

winning slot is fixed as the lifetime for the schedule.

3.12.1. Advantages of TRAMA

TRAMA achieves higher percentage of sleep time of the sensor

nodes and less collision as compared to many CSMA based

protocols. Substantial amount of energy is saved during commu-

nication through this protocol. Also, as the intended recipients of a

message are indicated with a bitmap, so less extent of commu-

nication is performed as compared to other protocols for multicast

and broadcast type of communication patterns.

3.12.2. Disadvantages of TRAMA

In their experiments, the authors (Rajendran et al., 2003) select

the transmission slots seven times longer than the signaling slots.

But, as each sensor node is either in receiving or transmitting state

during the random access period, so this consideration signifi-

cantly increases the duty cycle. Moreover, each node in TRMA

calculates the priorities of its two-hop neighbor for a time slot and

repeats these calculations for every time slot assignment which

leads to considerable energy consumption by the nodes.

3.13. Self-organizing MAC (SMACs)

SMACs (Sohrabi et al., 2000) is a hybrid MAC protocol for WSNs

which uses TDMA and Code Division Multiple Access (CDMA) for

accessing the communication channel and frequency hopping. It

utilizes a distributed mechanism which enables the sensor nodes

to discover their neighbors and communicate with them without






the help of any local or global master nodes. After the deployment

of the nodes in the network, each node wakes up randomly

according to some predefined distribution. It is assumed that the

nodes are able to turn their radios on or off and also they can tune

the carrier frequency to different bands. Unlike some earlier

methods, SMACs assigns a channel to a link immediately after

discovering the existence of the link. In this manner, nodes start

accumulating in a connected network, joined by concurrent links.

In order to reduce possible time collisions among slots between

adjacent links, distinct frequency bands are assigned to each link.

Once a link is established, a node has the knowledge about when

to turn on its radio for communicating. If no communications are

scheduled, it turns off its radio.

SMACs is particularly suitable for distributed sensor networks

in which there are a large number of static nodes with highly

constrained sources of energy. The ability of the nodes to discover

the neighbors and communicate with them leads to energy

efficiency, reduced implementation cost and high adaptability to

network topology changes. However, a downside of this protocol is

the increasing possibilities for collisions and lack of time synchro-

nization. Further, the allocated time slots are wasted if the sender

does not possess enough data to transmit to the recipient nodes

(Yadav et al., 2009).

3.14. Energy aware TDMA based MAC

It is a TDMA based MAC protocol for WSN whose primary goal

is to save energy consumption during network communication

(Arisha et al., 2002). In this protocol, different clusters of sensor

nodes in the network are formed and each node in the cluster is

managed by the gateway. The primary operations of a gateway are

to collect information about the sensor nodes within its own

cluster, allocate time slots to the constituent nodes, inform about

the time slots to all its intra-cluster nodes, perform fusion of data,

communicate with other gateways and finally transmit the data.

This protocol consists of four phases: data transfer, refresh, event-

triggered rerouting and refresh-based rerouting. Data is sent in the

allotted time slots during the data transfer phase. The sensor

nodes update their respective states, such as total energy level,

current position to their cluster's gateways in the refresh phase.

The gateways use the state information about the individual nodes

for transmitting data during event-triggered rerouting phase. The

refresh based rerouting periodically occurs after refreshing the

relevant information from the gateways and updating with new

information. The gateways perform the transmission and routing

during the two routing phases through executing the routing

algorithm.

The allocation of time slots is performed through graph parsing

strategy. Two approaches are proposed in this regard (Arisha et al.,

2002): Breadth First Search (BFS) and Depth First Search (DFS). BFS

assigns contiguous time slot numbers to the sensor nodes by

starting from the outermost node, whereas DFS assigns contiguous

time slots to the nodes on the route from outermost to the

gateway. The authors have performed experiments to analyze

the per-packet energy consumption, end-to-end delay, through-

put, etc. for both BFS and DFS based time slot allocation. BFS

technique saves considerable amount of energy which is con-

sumed by the sensor nodes in switching between ON/OFF states

and as such provides high lifetime of the sensor nodes. But, it

requires that the sensor nodes have sufficient buffer capacity and

as such it is prone to the buffer overflow problem. On the other

hand, DFS does not save the energy consumption by the sensor

nodes in switching between the ON/OFF states but avoids the

buffer overflow problem. However, it has low latency and high

throughput compared to BFS.

3.15. Berkeley media access control (B-MAC)

B-MAC (Polastre and Hill, 2004) is the most popular low

complexity contention based MAC protocol for WSNs which

provides a flexible interface to achieve ultra low-power network

operation, effective collision avoidance and high channel utiliza-

tion. B-MAC employs an adaptive preamble sampling mechanism

to reduce duty cycle and minimize idle listening so that signifi-

cantly low-power network operation can be performed. Duty cycle

reduction is achieved by using an asynchronous sleep/wake

scheme, viz. Low Power Listening (LPL) which is based on periodic

listening of the nodes. The sensor nodes wake up periodically to

sense the channel for network activity. The period between two

consecutive wake ups of a node is called the check interval,

whereas the period for which a node remains active after waking

up is called the wakeup time. Although, the wakeup time is fixed,

the check interval may be application specific. The packets in B-

MAC are made up of a long preamble and payload, where the

preamble duration is at least equal to the check interval so that

every node can detect an ongoing transmission in the network

during its check interval. A salient feature of this approach is that

it does not require synchronization of the nodes (Anastasi et al.,

2009; Polastre and Hill, 2004).

B-MAC also supports on trip reconfiguration and bidirectional

interfaces for system services in order to optimize performance.

Adaptability to changing traffic and network conditions and

scalability to large number of nodes are other major design goals

of B-MAC. The whole implementation scheme of B-MAC is main-

tained through an intelligently designed robust analytical model

(Polastre and Hill, 2004). Through comparing B-MAC and S-MAC,

the authors found that the flexible and adaptive characteristics of

B-MAC results in achieving significantly better performance of B-

MAC in terms of packet delivery rates, latency, throughput and

also often energy efficiency than S-MAC. These highly encouraging

findings, together with the implementation simplicity, low com-

plexity and flexible design of B-MAC surely make it one of the top

priority choices in many practical wireless sensor network

applications.

3.16. Data gathering MAC (D-MAC)

In many sensor network applications, it is often common that a

large part of the traffic consists of data which are assembled from

various sources to a particular sink via a unidirectional tree. This

type of most frequently observed tree based communication

pattern in sensor networks is known as convergecast type of

communication. D-MAC (Lu et al., 2007) is a schedule based

adaptive duty cycle MAC protocol which is specifically designed

and optimized for convergecast type of sensor network commu-

nication. The primary goal of D-MAC is to attain reasonably low

latency yet maintaining energy efficiency. The earlier protocols,

such as S-MAC and T-MAC suffer from a data forwarding inter-

ruption problemwhich is caused because not all nodes on a multi-

hop path to the destination can be notified about the ongoing data

transmission. This leads to significant sleep delay as well as limits

the data forwarding process to only a few hops (Anastasi et al.,

2009). Such undesired shortcomings of earlier protocol are solved

in D-MAC by staggering the active/sleep schedules of the sensor

nodes according to their position in the data gathering tree, as

depicted in Fig. 15. This scheme allows delivery of packets

continuously because all nodes in a multi-hop path can now be

notified about the ongoing transmission. Latency is minimized by

assigning adjacent slots to the successive nodes in the data

gathering tree. Duty cycle of each sensor node is also adaptively

adjusted according to the traffic load by varying the number of

active slots in a schedule interval.






In order to mitigate the problems of channel contention and

collisions, D-MAC further proposes a data prediction mechanism

andmore to send (MTS) packets. The data prediction scheme allows

every child in the multi-hop path to get chance for transmitting its

packets. MTS packets on the other hand resolve the interference

problem among nodes on different branches of the tree. A soft

timer is also maintained to handle lack of schedule synchroniza-

tion due to the accidental loss of the MTS packets.

3.16.1. Advantages of D-MAC

D-MAC is able to provide very good latency compared to other

sleep/listen based protocols and as such it can be a very good

choice for applications in which low latency is the primary

intention. The simulation results (Lu et al., 2007) show that D-

MAC significantly outperformed S-MAC in terms of energy effi-

ciency, latency and throughput for both multi-hop chain topology

and random data gathering tree topology.

3.16.2. Disadvantages of D-MAC

One major drawback of D-MAC is that the data transmission

paths among the sources and the sink node may be unknown

beforehand which makes it impossible to build up the data

gathering tree. Further, D-MAC is often prone to packet collisions

due to the lack of a robust collision avoidance mechanism.

3.17. Lightweight medium access protocol (LMAC)

LMAC (Van Hoesel and Havinga, 2004) is a lightweight and

highly energy efficient MAC protocol which adopts a low-

complexity mechanism for reserving the time slots. This protocol

achieves collision-free communications through TDMA, whereas

slot assignment and time synchronization through a self-

organizing mechanism. The primary goal of this protocol is to

minimize the radio state transitions and significantly reduce the

protocol overhead in the physical layer. Minimization of the radio

state switches makes the sensor nodes adaptive to the traffic load

and also limits the implementation cost. Nodes communicate

collision-freely by always transmitting messages, consisting of

two parts: control message and data unit during their allocated

time slots. A control message is of fixed length and serves multiple

purposes, e.g. it carries the address of its time slot controller,

indicates the distance of the associated node to the gateway,

addresses the intended receiver and reports the length of the data

unit. The timeout interval is kept short in order to ensure that the

sensor nodes do not waste precious energy through idle listening.

A notable shortcoming of LMAC is its usage of fixed length frames,

where the length of the frames must have to be specified before

nodes deployment which is often difficult and may not be

practical.

LMAC is broadly inspired by the EMACs protocol. Like EMACs,

the network operations in LMAC are not controlled by a central

manger or base station and the nodes can choose their own time

slots on the basis of local information only. The obtained simula-

tion results (Van Hoesel and Havinga, 2004) show that LMAC was

able to prolong the overall network lifetime by a factor of 2.4 and

3.8, as compared to EMACs and S-MAC respectively.

3.18. Pattern MAC (PMAC)

PMAC (Zheng et al., 2005) is a time slotted MAC protocol,

primarily designed to minimize energy waste from idle listening

which is common in some earlier protocols, e.g. S-MAC and T-

MAC. Many MAC protocols face the challenge of reducing idle

listening which is a major source of energy consumption. In

S-MAC, the sensor nodes wake up periodically to check if an

active transmission is in progress even if there is none. This

problem can be somewhat reduced by keeping a small duty cycle

but this in turn degrades the network throughput under heavy

traffic load. Similar case also happens in T-MAC in which the nodes

have to wake up at the beginning of each frame and remain awake

for at least the timeout period, even when there is no network

activity in the neighborhood.

PMAC attempts to overcome this drawback of earlier protocols

by allowing a sensor node to acquire knowledge about its

neighborhood before hand through a sleep/wake-up pattern. A

pattern for a sensor node is a string of bits which indicates its

tentative sleep/wake-up plan over several time slots and is also

adaptively changed according to the network traffic. The patterns

for all the nodes are generated through a robust mathematical

mechanism. For performing pattern exchange among the nodes,

time is divided into a number of Super Time Frames (STFs), where

each STF consists of two sub-frames: Pattern Repeat Time Frame

(PRTF) and Pattern Exchange Time Frame (PETF). The nodes repeat

their respective patterns during PRTF, whereas new patterns are

exchanged among the neighboring nodes during PETF. The num-

ber of time slots in PETF is set equal to the maximum possible

number of neighbors of a sensor node. This design of PMAC

requires some amount of time synchronization among the sensor

nodes, however it need not to be very strict.

The underlying mechanism of PMAC makes it to considerably

reduce energy consumption from idle listening by permitting the

sensor nodes, not involved in any network activity to remain

asleep. In order to further conserve energy, the timeout scheme of

T-MAC can also be additionally introduced. The empirical results

(Zheng et al., 2005) demonstrated that in comparison to S-MAC,

PMAC is able to achieve more power conservation under light

traffic loads and higher throughput under heavier traffic loads.

One drawback with PMAC is that it requires sleep/wake-up

patterns to be announced before hand which is often difficult

and not feasible in a practical sensor network application.

3.19. Zebra MAC (Z-MAC)

Z-MAC (Rhee et al., 2005) is a hybrid MAC protocol for WSNs

which combines both TDMA and CSMA in order to achieve high

channel utilization, improved medium contention, low latency and

collision avoidance at a low cost. It acts like CSMA or TDMA under

low and high contention levels, respectively. Z-MAC assigns time

slots to the sensor nodes at the time of deployment of nodes.

Although this scheme incurs high initial overhead, but it is

eventually compensated by enhanced throughput and energy

efficiency. Unlike the traditional TDMA, a node may transmit

during any time slot after performing adequate carrier sensing.

Fig. 15. Data gathering tree in D-MAC implementation.






The owners of a particular time slot always get higher priority in

accessing the channel than the non-owners. This regulation

effectively reduces the chances of collisions during data transmis-

sions. By combining CSMA and TDMA, Z-MAC becomes more

preventive against some of the network hazards which include

timing failures, time wise variation in channel conditions, slot

assignment failures and topology changes.

Z-MAC starts with a preliminary setup phase in which the

following four operations are performed sequentially (Anastasi

et al., 2009; Rhee et al., 2005): neighbor discovery, slot assignment,

local frame exchange and global time synchronization. Each node

builds a list of its two-hop neighbors in the neighbor discovery

phase and then gets its respective time slot through a distributed

slot assignment algorithm. The assignment of slots to the nodes is

performed in a collision-free manner which ensures that no two

nodes in the two-hop neighborhood are allocated the same slot.

Then, in the local frame exchange phase, each node decides its

period of time, known as the time frame for which it can perform

network operations in the allocated time slot. The time frame of a

node in Z-MAC is selected locally to fit its local neighborhood size

but avoiding any conflicts with its contending neighbors. In the

global time synchronization step, the time frame and local slot

assignment of each node is forwarded to its two-hop neighbors.

Thus, all nodes achieve the slot and frame information about their

two-hop neighbors and accordingly synchronize to a common

reference time slot. After the setup phase, nodes access the

channel through the transmission control regulation, according to

which each node can be in one of the levels: Low Contention Level

(LCL) or High Contention Level (HCL). Usually a node is in LCL, except

when it receives an Explicit Contention Notification (ECN) message

from a two-hop neighbor within the last TECN period. These ECN

messages are sent by the sensor nodes which experience high

contention. In LCL, any node can contend for the channel, whereas

in HCL, only the owners of the current slot and their one-hop

neighbors are allowed for contention. Although, in both levels,

owners have higher priority than non-owners in accessing the

channel, but a non-owner can steal the slot when the slot does not

have a owner or when its owner have nothing to transmit. In this

way, high channel utilization is achieved even under low contention.

As specified in the outset, Z-MAC has many advantages over

traditional non-hybrid protocols, such as S-MAC in terms of channel

utilization, latency, throughput and energy efficiency. The dynamic

adaptation to contention level is another salient feature of this

protocol. It has been observed that even in the worst case, Z-MAC

performs as good as the traditional CSMA (Rhee et al., 2005).

3.20. X-MAC

X-MAC (Buettner et al., 2006) is a low-power, energy efficient

MAC protocol which is specifically designed for asynchronous

duty-cycled WSNs. Traditional MAC protocols for duty-cycled

sensor networks, such as B-MAC and WiseMAC adopt preamble

sampling with an extended length of the preamble. In spite of

many advantages, the use of long preamble length in low-power

listening suffers from increased per-hop latency, idle listening by

the senders and recipients and energy waste from overhearing by

the non-intended recipients. X-MAC aims to overcome these

drawbacks through employing a shortened preamble approach

yet retaining the salient features of low-power listening, e.g.

implementation simplicity, reduced power consumption and

decoupling of sender and transmitter sleep schedules. This proto-

col adopts a short strobed preamble which allows the target

receiver to interrupt the long preamble immediately after it wakes

up and finds that it is the target. Doing this saves a lot of time and

energy which are otherwise wasted in waiting for the entire long

preamble to complete. X-MAC also significantly reduce per-hop

latency and energy consumption by indicating the address of the

intended recipient in its short preamble, so that non-intended

recipients can quickly go back to sleep. Further, an adaptive

algorithm is suggested which dynamically adjusts the duty cycle

of the receivers in order to optimize per-packet latency or energy

consumption or both.

The empirical results (Buettner et al., 2006) demonstrated that

the shortened preamble approach of X-MAC indeed achieved

significantly reduced energy consumption at the senders and

receivers end, reduced per-hop latency and increased throughput.

Moreover, X-MAC was found to be adaptable to various conditions

of network traffic.

3.21. Funneling-MAC

The funneling-MAC (Ahn et al., 2006) is a hybrid MAC protocol

which is specifically designed to mitigate a unique undesirable

behavior of sensor networks, commonly known as the funneling

effect. In a static sensor network, packets coming from the sensor

nodes follow a many-to-one multi-hop path towards one or more

sink points and this travel pattern together with the centralized

data collection at a sink leads to dramatic congestion of packets

(choke point) near the intended sink node (Anastasi et al., 2009; Ahn

et al., 2006). This unique behavior of a sensor network is known as

the funneling effect and it has various adverse consequences on the

communication flow of the network. The funneling effect signifi-

cantly increases transit traffic intensity, delay in packets delivery,

congestion, occurrences of collision, loss of packets and energy

wastes when the packets move closer to the sink node. Moreover,

the funneling of packets leads to considerably limited application

fidelity at the sink nodes. A diagrammatic illustration of the

funneling effect, due to Ahn et al. (2006) is depicted in Fig. 16.

Alleviating the funneling effect is a major challenge for sensor

networks researchers and till now, various approaches have been

suggested for this purpose (Hull et al., 2004; Shrivastava et al.,

2004; Wan et al., 2005). However, most of these methods attempts

to control network traffic load and congestion at the aggregation

points or source nodes which is often very difficult. As such, Hull

et al. (2004) and Wan et al. (2005) have rightly pointed out that

none of these proposed techniques alone can fully alleviate the

problem of funneling. Some popular MAC protocols, such as

S-MAC, T-MAC, B-MAC, TRAMA and Z-MAC attempt to diminish

the funneling effect but only to limited extents.

The funneling-MAC (Ahn et al., 2006) is a localized, sink-

oriented MAC protocol which is designed to explicitly recognize

as well as mitigate the funneling effect in a sensor network. It

adopts a hybrid mechanism which implements a CSMA/CA

throughout the network, whereas a localized TDMA in the funnel-

ing or high intensity region. In funneling-MAC, the tasks of

Fig. 16. The funneling effect in a sensor network.






assessing and maintaining the depth of the intensity region as well

as TDMA scheduling of sensor events in that region are all

performed by the sink node only. The funneling-MAC assumes

that the sink node is apparently more energy rich and computa-

tionally efficient than the non-sink sensor nodes, although this

assumption may be atypical in some practical applications. In view

of the mentioned characteristics, it is evident that this MAC

protocol puts most of the network-related workloads on the sink

node and as such it can be viewed as a sink-oriented protocol. The

sink-oriented nature, together with the use of a localized TDMA in

the funneling region significantly helps this MAC protocol to boost

application fidelity in the sensor network as well as to mitigate the

funneling effect and scalability problems. In a nutshell, the

funneling-MAC attempts to eliminate the funneling effect and

improve network performance through putting additional control

over the first few hops of the network from the sink node (Ahn

et al., 2006).

The funneling-MAC is based on an effective hybrid mechanism

and benefits from the region-wise implementations of CSMA/CA

and TDMA. The empirical results (Ahn et al., 2006) demonstrate

that this protocol is energy efficient, notably improves throughput,

controls packets losses, boosts fidelity in the sensor network and

significantly outperforms the earlier B-MAC and Z-MAC protocols

in terms of eradicating the funneling effect under a wide variety of

traffic conditions. In the downside of this MAC protocol, there are

high implementation cost, putting excessive burdens on the sink

nodes and difficulty in precisely identifying the zone of nodes

which should gain additional control.

4. Discussions and future research directions

In the previous section, a wide variety of MAC protocols for

WSNs are discussed together with their associated merits and

demerits. Among them, some protocols, such as MACA, MACAW,

S-MAC, and T-MAC are described with more details than others.

This is because these protocols are highly important and moti-

vated the development of many later MAC protocols. For example,

S-MAC can be regarded as a benchmark as it alone inspired several

other MAC protocols, e.g. T-MAC, DS-MAC, and optimized MAC.

Further, many authors often prefer to evaluate the performances of

their proposed MAC protocols against those of S-MAC. It should

also be noted that MAC protocols for WSNs is a very dynamic area

with nonstop contributions from many researchers. There is an

ongoing trend of developing new MAC protocols from different

novel perspectives and so the total count of the available MAC

protocols is increasing persistently. Considering these facts, the

preceding section carries out an intensive study of the various

important MAC protocols which are developed till now for WSNs;

but, the study is in no way exhaustive. This section is devoted to

provide a brief comparison of the MAC protocols which are

described so far and outline a number of future research directions

in this domain. All the discussed MAC protocols are concisely

presented, together with their salient characteristics in Table 1.

From Table 1, it can be seen that energy conservation is the

primary goal of all the MAC protocols which are proposed so far.

Conservation of energy is achieved through adopting various means

which include dual communication channels, preamble sampling,

adaptive duty cycle, clustering of sensor nodes and sleep–wake

patterns. Furthermore, most of the developed MAC protocols sup-

port all types of communication patterns and are also quite good in

adapting to the changes in network topology and size. Another

important fact is that in spite of so many existing MAC protocols for

WSNs, none can be accepted as the absolute standard. The choice of

a MAC protocol is normally based on the nature of the sensor

network, specific performance goals and implementation cost. Now,

we isolate a number of open issues and research scopes which will

be helpful in designing future MAC protocols for WSNs.

The present MAC protocols are mostly based on either CSMA or

TDMA methods of medium utilization. Traditional CSMA methods

Table 1

A concise comparison of the studied MAC protocols.

Protocol Category Type Comm.

pattern

Adaptability to

changes

Salient features

MACA, MACAW, IEEE 802.11 Contention

based

CSMA All Good Simple to implement, energy efficient, resolve the problem of

hidden terminal

PAMAS Contention

based

CSMA All Good Uses two communication channels, energy efficient with increased

design cost

S-MAC, Contention

based

CSMA All Good Highly energy efficient, loose time synchronization requirements

T-MAC,

DS-MAC

EMACs, LMAC Schedule

based

TDMA All Good Easy to implement, flexible, avoids collisions and highly energy

efficient

WiseMAC, B-MAC, X-MAC Contention

based

CSMA/np-CSMA All Good Employ preamble sampling to reduce duty cycle and idle listening

and also avoid collisions

Sift Contention

based

CSMA/CA All Good Designed for event-driven sensor networks, achieves reasonably

low latency with large network size

Optimized MAC Contention

based

CSMA All Good Uses adaptive duty cycle and minimizes control packet overhead

TRAMA Schedule

based

TDMA All Good Higher sleep times for nodes and less collisions

SMACs Hybrid TDMA and CDMA Distributed

WSNs

Very good Energy efficient, low cost, but prone to collisions and

synchronization problems

Energy Aware MAC Schedule

based

TDMA All Good Forms clusters of sensor nodes which are managed by gateways to

save energy

D-MAC Schedule

based

TDMA Convergecast Weak Uses adaptive duty cycle to attain low latency and high energy

efficiency

PMAC Schedule

based

TDMA All Moderate Uses a sleep–wake pattern to primarily reduce energy wastes from

idle listening

Z-MAC Hybrid TDMA and CSMA All Very good Achieves low latency, high channel utilization and collision

avoidance at a considerably low cost.

Funneling-MAC Hybrid Localized TDMA and

CSMA/CA

All Very good Specifically designed to mitigate the highly adverse funneling effect

in WSNs






have the advantages of lower delay and increased throughput at

lower traffics but they cannot adequately handle the event-driven

workloads which are very important in practical WSN applica-

tions. Also, CSMA lacks well-defined collision detection and

avoidance schemes. More research works are required to over-

come these drawbacks of the traditional CSMA. One notable

approach in this direction is the CSMA/p* scheme (Tay et al.,

2004) which uses np-CSMA with the unique non-uniform prob-

ability distribution p* in order to minimize collisions between

contending stations. Contrary to CSMA, TDMA performs collision-

free medium access but it has other inherent limitations which

include degraded throughput and idle slots at low traffic, lack of

nodes synchronization and difficulty to cope with topology

changes. Till now, only a few MAC protocols have been designed

which attempt to reduce the drawbacks of TDMA. For example,

EMACs and LMAC considerably enhance the performances of

TDMA by allowing the sensor nodes to choose their own time

slots locally without depending on a central manager or base

station. From the foregoing discussion, it is evident that a hybrid

MAC protocol which combines CSMA and TDMA can benefit from

the merits of both the schemes while mitigating their drawbacks.

A few MAC protocols which adopt this hybridization approach are

proposed in literature but this topic surely needs further explora-

tions (Demirkol et al., 2006). Apart from CSMA and TDMA, other

potential schemes such as CDMA and Frequency Division Multiple

Access (FDMA) which received least attentions so far in MAC

protocols designing should also be considered. Performance eva-

luation of different medium access mechanisms in a common

framework is another major issue which is missing in the present

works.

Inter-protocol combination may be a new research direction in

which the methodologies of two or more MAC protocols can be

effectively combined, unless any conflict occurs. Although, this

practice may be a little costly at times but it can substantially

improve the network operations as well as diminish various

power-related constraints of the sensor nodes. For example, the

concepts of adaptive preamble sampling of B-MAC and distributed

communication of SMACs may be combined to design a new MAC

protocol which is apparently collision-free, requires less duty

cycle, highly energy efficient and easily adaptable to network

topology changes.

Most of the available MAC protocols are designed with the

primary goal of minimizing energy waste which of course is a

crucial challenge for WSN applications. However, there are some

other important issues to be considered as well which include

network security, mobility of nodes, reliability of packets delivery

and protocol evaluation on real sensor platforms (Yadav et al.,

2009). Sensor networks are prone to a number of malicious

attacks, such as eavesdropping, node capturing, physical temper-

ing and denial of service. Protecting against such type of attacks is

quite challenging, mainly because the security techniques of

traditional networks cannot be directly applied to sensor network

systems (Karlof et al., 2004). In spite of the paramount importance

of maintaining sensor network security, only limited research has

been carried out in this regard. The first fully implemented link

layer security architecture, viz. TinySec has been proposed by

Karlof et al. (Perrig et al., 2004) only a few years ago. Evidently

more in depth works are required in order to develop advanced

sensor networks security mechanisms.

Mobility of sensor nodes is a very useful concept in which

increasing research interests have been observed in recent years.

Traditional static sensor networks often suffer from severe energy

consumption problems, such as the funnelling effect. Introducing

mobility to some (if not all) sensor nodes can make the WSN

significantly energy efficient and suitable for various practical

dynamic scenarios (Yadav et al., 2009; Anastasi et al., 2009).

However, this domain is in its initial phase and many related

aspects have to be carefully studied. There are excellent future

scopes for designing energy-efficient mobility-supported MAC

protocols for WSNs within the range of affordable costs. Reliable

delivery of packets in practical sensor network applications is

another promising area which has to be extensively explored in

future works. Moreover, the MAC protocols should be tested on

real WSN platforms, rather than using only simulated architec-

tures. This will provide a fair idea of the performances of a

proposed protocol in real life scenarios (Yadav et al., 2009).

Integration of MAC with other routing layers has several

benefits and as such it is an intriguing area, but has limited

research attention so far. Integrating different layers increases

cross-layer interaction which in turn enhances energy saving

through minimizing layer-wise packet overheads and also allows

more efficient operation of the sensor network (Demirkol et al.,

2006). Relying upon only a single layer can misinform about the

overall performance of the sensor network which can adversely

affect the evaluation of a proposed MAC protocol for the network.

Although, there are some recognized works (Cui et al., 2005; Zorzi,

2004; Ding et al., 2003) aiming to integrate MAC with other layers

but more intensive research has to be carried out in future.

5. Conclusions

Wireless sensor networks comprise an emerging technology

which has found overwhelming applications in a wide range of

practical communication scenarios during the last decade. A

typical WSN is composed of several battery-operated, usually

static sensor nodes which are constrained in energy supply,

bandwidth, storage capacity and processing ability. The primary

objective of a MAC protocol is to enable smooth operation of the

associated WSN and prevent energy consumption from all poten-

tial sources. Although, the domain of MAC protocols is relatively

new, it is rapidly enriching in both quality and quantity through

active contributions from research community.

In this paper, at first the designing challenges of MAC protocols

for WSNs are discussed and then a wide variety of existing MAC

protocols are meticulously studied, highlighting their inherent

merits and demerits. In order to provide an up-to-date survey,

several recently developed MAC protocols, e.g. PMAC, X-MAC, Z-

MAC are described together with the earlier well recognized

protocols, e.g. MACA, MACAW, IEEE 802.11, S-MAC. The important

findings of the present study are summarized as follows. First, it is

observed that in spite of their abundance in literature, the choice

of a MAC protocol for a particular WSN application is not

straightforward and is normally problem-dependent. Also, none

of the designed MAC protocol can be selected as the standard one

for all sensor networking scenarios. Second, although, energy-

efficiency is the fundamental objective, a MAC protocol can seldom

prevent energy wastes from all potential sources. Third, some

protocols, e.g. MACA, S-MAC, T-MAC have significantly inspired the

designs of various other MAC protocols. The penultimate section of

this paper concisely compares all the studied MAC protocols and

discusses some important future research directions in this

domain. Although, the present study is quite intensive, but is of

course not exhaustive. This paper includes several important MAC

protocols, designed so far in sensor networking literature, but not

all. The counts of available MAC protocols are continuously

increasing and as such time-to-time surveys are very important.

Hopefully, the present study will be significantly helpful in under-

standing the current research trend in designing MAC protocols

for WSNs.






References

Ahn GS, Hong SG, Miluzzo E, Campbell AT, Cuomo F. Funneling-MAC: A localized,

sink-oriented MAC for boosting fidelity in sensor networks. In: Proceedings of

the 4th ACM international conference on embedded networked sensor systems

(SenSys); October 2006. pp. 293–306.Akkaya K, Younis M. A survey on routing protocols for wireless sensor networks.

Ad hoc Netw 2005;3(3):325–49.Akyildiz I, Su W, Sankarasubramaniam Y, Cayirci E. A survey on sensor networks.

IEEE CommunMag 2002:102–14.Anastasi G, Conti M, Di Francesco M, Passarella A. Energy conservation in wireless

sensor networks: a survey. Ad Hoc Netw 2009;7(3):537–68.Arisha K, Youssef M, Younis M. Energy aware TDMA based MAC for sensor network.

In: Proceedings of the IEEE workshop on integrated management of power

aware communications computing and networking (impacct); 2002.Bharghavan V, Demers A, Shenker S, Zhang L. MACAW: a media access protocol for

wireless LAN's. ACM SIGCOMM ComputCommun Rev 1994;24(4):212–25.Buettner M, Yee GV, Anderson, E, Han R. X-MAC: a short preamble mac protocol for

duty-cycled wireless sensor networks, In: Proceedings of the 4th ACM inter-

national conference on embedded networked sensor systems (SenSys); October

2006, pp. 307–320.Calí G, Ghizdavescu I, Grauballe A, Jensen MG, Pozzo F. MAC protocol for wireless

sensor network, Project Report. Communications Systems, Group 652, Aalborg

University; 2007.Cui S, Madan R, Goldsmith A, Lall S. Joint routing, MAC, and link layer optimization

in sensor networks with energy constraints. In: Proceedings of the IEEE

international conference on communications (ICC), Seoul, Korea; May 16–20,

2005. pp. 725–9.Demirkol I, Ersoy C, Alagöz F. MAC protocols for wireless sensor networks: a survey.

IEEE Commun Mag 2006;44(4):115–21.Ding J, Sivalingam K, Kashyapa R, Chuan LJ. A multi-layered architecture and

protocols for large-scale wireless sensor networks. In: IEEE 58th vehicular

technology conference (VTC 2003-Fall), Orlando, Florida, USA; October 6–9,

2003. pp. 1443–7.El-Hoiydi A. Spatial TDMA and CSMA with preamble sampling for low power ad-

hoc wireless sensor network. In: Proceedings of the seventh international

symposium on computers and communications (ISCC); July 2004. pp. 685–92.Enz CC, El-Hoiydi A, Decotignie JD, Peiris V. WiseNET: an ultralow-power wireless

sensor network solution. IEEE Comput 2004;37(8):62–70.Haartsen JC. The bluetooth radio system. IEEE Pers Commun 2000:28–36.Heinzelman WR, Chandrakasan A, Balakrishnan H. Energy-efficient communication

protocol for wireless microsensor networks. In: Proceedings of the 33rd IEEE

annual Hawaii international conference on system sciences; January 4–7, 2000.

pp. 1–10.Hull B, Jamieson K, Balakrishnan H. Mitigating congestion in wireless sensor

networks. In: Proceedings of the 2nd ACM international conference on

embedded networked sensor systems (SenSys); November 3–5, 2004.

pp. 134–47.Jamieson K, Balakrishnan H, Tay YC. Sift: A MAC protocol for event-driven wireless

sensor networks. In: Proceedings of the European workshop on wireless sensor

networks (EWSN); February 13–15, 2006, pp. 260–275.Karlof C, Sastry N, Wagner D, TinySec: a link layer security architecture for wireless

sensor networks. In: Proceedings of the 2nd ACM international conference on

embedded networked sensor systems (SenSys); November 3–5, 2004. pp. 162–

75.Karn P. MACA–a new channel access method for packet radio. In: Proceedings of

the ARRL/CRRL amateur radio 9th computer networking conference; September

22, 1990.

Lin P, Qiao C, Wang X. Medium access control with a dynamic duty cycle for sensornetworks. In: Proceedings of the IEEE wireless communications and network-ing conference (WCNC); March 2004. pp. 1534–9.

Lu G, Krishnamachari B, Raghavendra CS. An adaptive energy-efficient and low-latency MAC for tree-based data gathering in sensor networks. WirelessCommun Mob Comput 2007;7(7):863–75.

Perrig A, Stankovic J, Wagner D. Security in wireless sensor networks. CommunACM 2004;47(6):53–7.

Polastre J, Hill J, Culler D. Versatile low power media access for sensor networks. In:Proceedings of the 2nd ACM international conference on embedded networkedsensor systems (SenSys), November 3–5, 2004. pp. 95–107.

Rajendran V, Obraczka K, Garcia-Luna-Aceves JJ. Energy-efficient, collision-freemedium access control for wireless sensor networks. In: Proceedings of the 1stACM international conference on embedded networked sensor systems(SenSys), Los Angeles, California, November 5–7, 2003. pp. 181–92.

Rhee I, Warrier A, Aia M, Min J. Z-MAC: a hybrid MAC for wreless sensor networks.In: Proceedings of the 3rd ACM international conference on embeddednetworked sensor systems (SenSys), Sun Diego, USA; November 2005.

Shrivastava N, Buragohain C, Agrawal D, Suri S. Medians and beyond: newaggregation techniques for sensor networks. In: Proceedings of the 2nd ACMinternational conference on embedded networked sensor systems (SenSys),November 3–5, 2004. pp. 239–249.

Singh S, Raghavendra CS. PAMAS: power aware multi-access protocol with signal-ing for ad hoc networks. Comput Commun Rev 1998;28(3):5–26.

Sohrabi K, Gao J, Ailawadhi V, Pottie GJ. Protocols for self-organization of a wirelesssensor network. IEEE Pers Commun 2000;7(5):16–27.

Stemm M, Katz RH. Measuring and reducing energy consumption of networkinterfaces in hand-held devices. IEICE Trans Commun 1997;E80-B(8):1125–31.

Tay YC, Jamieson K, Balakrishnan H. Collision-minimizing CSMA and its applicationsto wireless sensor networks. IEEE J Selected Areas Commun 2004;22(6):1048–57.

Tilak S, Abu-Ghazaleh NB, Heinzelman W. A taxonomy of wireless micro-sensornetwork models. ACM SIGMOBILE Mob Comput Commun Rev 2002;6(2):28–36.

Van Dam T, Langendoen K. An adaptive energy efficient MAC protocol for wirelessnetworks. In: Proceedings of the 1st ACM conference on embedded networkedsensor systems; 2003.

Van Hoesel LF, Havinga PJM. A lightweight medium access protocol (LMAC) forwireless sensor networks: reducing preamble transmissions and transceiverstate switches. In: Proceedings of the 1st international workshop on networkedsensing systems (INSS), Tokyo, Japan; 2004.

Van Hoesel LFW, Nieberg T, Kip HJ, Havinga PJM. Advantages of a TDMA based,energy-efficient, self-organizing MAC Protocol for WSNs. In: Proceedings of theIEEE vehicular technology conference (VTC), May 2004. pp. 1598–602.

Wan CY, Eisenman SB, Campbell AT, Crowcroft J. Siphon: overload traffic manage-ment using multi-radio virtual sinks in sensor networks. In: Proceedings of the3rd ACM international conference on embedded networked sensor systems(SenSys), Sun Diego, USA; November 2005. pp. 116–29.

Yadav R, Varma S, Malaviya N. Optimized medium access control for wirelesssensor network. Int J Comput Sci Netw Secur 2008;8(2):334–8.

Yadav R, Varma S, Malaviya N. A survey of MAC protocols for wireless sensornetworks. UbiCC J 2009;4(3):827–33.

Ye W, Heidemann J, Estrin D. An energy-efficient MAC protocol for wireless sensornetworks. Proc IEEE INFOCOM 2002:1567–76.

Zheng T, Radhakrishnan S, Sarangan V. PMAC: an adaptive energy-efficient macprotocol for wireless sensor networks. In: Proceedings of the 19th IEEE paralleland distributed processing symposium (IPDPS); 2005. pp. 65–72.

Zorzi M. A new contention-based mac protocol for geographic forwarding in ad hocand sensor networks. In: Proceedings of the IEEE international conference oncommunications (ICC), Paris, France; June 20–24, 2004. pp. 3481–5.



http://refhub.elsevier.com/S1084-8045(14)00024-1/sbref1







































0 study of various mac protocols for wsn

Documents