november 4, 2003apoc 2003 wuhan, china 1/14 demand based bandwidth assignment mac protocol for...
TRANSCRIPT
November 4, 2003 APOC 2003Wuhan, China
1/14
Demand Based Bandwidth Assignment MAC Protocol for
Wireless LANs
Presented by Ruibiao QiuDepartment of Computer Science and Engineering,
Washington University
St. Louis, MO 63130, USA
K. Murugan, B. Dushyanth, E. Gunasekaran, S. Arivuthokai, R. S. Bhuvaneswaran, S. Shanmugavel
Department of Computer Science, CEG, College of Engineering
Anna University, Chennai, India
November 4, 2003 APOC 2003Wuhan, China
2/14
INTRODUCTION• IEEE 802.11
– Up to 54Mbps of raw physical data. – Uses Aloha protocol to share the wireless medium. – Drawback:
• cannot work under high traffic load.
• Wireless medium is highly bandwidth and power limited– A TDMA solution may increase channel bandwidth
utilization– TDMA major disadvantage
• Fixed time slots could limit the number of nodes.
November 4, 2003 APOC 2003Wuhan, China
3/14
INTRODUCTION
• DAMA-TDMA – Used by many satellite networks where the
time slots are allocated dynamically.
• We propose a variation of DAMA-TDMA– Demand Based Bandwidth Assignment
(DBBA) – Challenge
• Avoid collision during the demand request contention period.
November 4, 2003 APOC 2003Wuhan, China
4/14
IEEE 802.11 MAC PROTOCOL
• CSMA/CA scheme.
• MAC layer depends on acknowledgement packet to determine collision. – Receiver
• Sends an acknowledge before stipulated time interval after completion of the transmission.
– Transmitter• If no acknowledgement within the timeout period,
assumes that the packet loss and retransmits
November 4, 2003 APOC 2003Wuhan, China
5/14
IEEE 802.11 MAC PROTOCOL
• Low bandwidth utilization due to MAC layer constraints.
• Number of nodes increases– Collision probability increases– Overall network bandwidth lower
• The latency depends on traffic conditions with no bandwidth guarantee.
November 4, 2003 APOC 2003Wuhan, China
6/14
MAC LAYER
November 4, 2003 APOC 2003Wuhan, China
7/14
DBBA-MAC PROTOCOL
• Two components in a network– A primary controller node – Multiple station nodes
• DBBA frame has n+3 time slots– Beacon time slot – Demand request time slot– Demand Assignment time slot– Data Time slots.
November 4, 2003 APOC 2003Wuhan, China
8/14
DBBA-MAC PROTOCOL• Controller
– Broadcasts a beacon packet during the Beacon time slot at the beginning of each frame.
– The beacon frame contains• network SSID, frame information, timeslot information.
• Station node– Sends a demand request packet to the controller during the Demand
Request time slot– Contains
• Source node ID, destination node ID, date size, and QoS requirement.• Controller
– Processes all the requests– Generates the bandwidth assignment table.– Broadcasts acknowledgement to all accepted requests in the Demand
Assignment time slot
November 4, 2003 APOC 2003Wuhan, China
9/14
DBBA NETWORK SIMULATION MODEL
November 4, 2003 APOC 2003Wuhan, China
10/14
SIMULATION• Simulation Model:
– Three classes: scheduler, modem, packet generator• Scheduler
– Collects the desired transmission time from all stations– Determines the modem with the lowest desired transmission time stamp– Detects and notifies collision
• Modem– Depicts the behavior of stations with information from the packet
generator.– Returns the packet details to scheduler.– Waits for acknowledge of the transmitted packet from the scheduler
• Packet Generator– Depicts a user generating traffic data. – One instance per Modem. – The uniformly random distributed packet length with three packets types.
• Small, medium, large.
November 4, 2003 APOC 2003Wuhan, China
11/14
No of Nodes vs Throughput (Traffic Load = 0.8)
0102030405060708090
100
3 6 9 12 15 18 21 24 27 30No of Nodes
Thro
ughp
ut (%
)
DDBA
IEEE802_11
No of Nodes vs Collisions (Traffic Load = 0.8)
0102030405060708090
100
3 6 9 12 15 18 21 24 27 30No of Nodes
Collis
ions
DBBA
IEEE802_11
SIMULATION RESULT
• As the number of nodes increases, the throughput of DBBA slightly increases due to higher traffic loads.
• The number of collisions is 5 % to 8% lesser than 802.11 as the number of nodes increases.
November 4, 2003 APOC 2003Wuhan, China
12/14
Traffic Load vs Throughput (No of Nodes = 100)
0102030405060708090
100
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Traffic Load
Thro
ughp
ut (%
)
DBBA
IEEE802_11
Traffic Load vs Collisions (No of Nodes = 20)
0102030405060708090
100
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Traff ic Load
Collis
ions
(%)
DBBA
IEEE802_11
SIMULATION RESULT
• Throughput and collisions of both 802.11 and DBBA are compared with different traffic loads.
• Throughput of DBBA increases as traffic load increases
• Amount of collisions is 5% to 8% lesse than 802.11 as the traffic load increases.
November 4, 2003 APOC 2003Wuhan, China
13/14
No of Nodes vs Average Latency (Traffic Load = 0.8)
0800
160024003200400048005600640072008000
3 6 9 12 15 18 21 24 27 30
No of Nodes
Avg
. Lat
ency
(US)
DBBA
IEEE802_11
Traffic Load vs Average Latency (No of Nodes = 20)
0500
100015002000250030003500400045005000
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Traffic Load
Avg
. Lat
ency
(US)
DBBA
IEEE802_11
SIMULATION RESULT
• Average latency of DBBA and 802.11 is compared with increasing number of nodes and different traffic loads.
• Delay is little more than the 802.11 as the number of nodes increases with traffic load 0.8
• Under different traffic load conditions, the average latency of DBBA protocols is higher.
November 4, 2003 APOC 2003Wuhan, China
14/14
CONCLUSION
• Demand Based Bandwidth Assignment (DBBA) protocol is proposed.– Higher bandwidth utilization– Larger number of stations in one cell– Less propagation delay– More power efficiency– Provide better quality of service.
• Efficiency increased without affecting the overall throughput
• Future work– Allocation of dedicated slots for a complete session to
reduce the delay for the constant bit rate services.