emerging radio …and the world

Emerging Radio …and the World

Aggelos Bletsas

Department of Electronic and Computer Engineering,Technical University of Crete (TUC), Greece

TUC Telecom Seminars 2008-2009

June 1, 2009

Collaborators: Prof. Sahalos, Prof. Win, Prof. Lippman & Dr. Dimitriou

June 1, 2009 TUC Telecom Seminars 2008-2009 2

Outline

1. RFIDs

2. Emergency Radio

3. Emerging Relaying in Wireless Access Networks

4. Education and integration


Backscatter Radio/RFIDs are used everywhere!


Brief Intro: Basic Principle

Reader

Tag

Communication from Tag to Reader = Backscatter Radio = = binary modulation on tag antenna-tag load reflection coefficient!

G. Vannucci, A. Bletsas and D. Leigh, "A Software-Defined Radio System for Backscatter Sensor Networks", IEEE Transactions on Wireless Communications (TWC), Vol. 7, No. 6, pp. 2170-2179, June 2008.


Problem Formulation

Bit ‘0’ ( )Z1ZL= Bit ‘1’ ( )Z2ZL=

Carrier

Backscattered Signal from Tag to Reader

Z1 Z2

Reader

Tag

Focus on Tag-to-Reader Communication

For given tag antenna, what are the ``optimal’’ tag loads Z1

(for bit ‘0’), Z2 (for bit ‘1’)?


Prior Art: focus on minimum scattering antennas only

AC ZL

Tag

ZL

Equivalence is true only for minimum scattering antennas

Za

Z1 Z2

Reader

Tag

[Nikitin et al Electronics Letters 2007]: “Since the scattered field is proportional to the complex current in the antenna…”, the system designer “should maximize |Γ1-Γ2|2…”(where Γi is the reflection coefficient for load Zi)

As stated by Nikitin et al, this is only true for minimum scattering antennas.

We derive the solution for the general tag antenna case.


Basic Quantities

Bit ‘0’ ( )Z1ZL= Bit ‘1’ ( )Z2ZL=

Carrier


E0: Antenna specific…

As: Tag Antenna

Structural Mode Parameter

σi: Radar Cross Section

(RCS)

at load ZL=Zi

AC ZL

Tag

ZL

Equivalence is true only for minimum scattering antennas

Za


``Optimality’’ Definition: Constraint 1

Bit ‘0’ ( )Z1ZL= Bit ‘1’ ( )Z2ZL=

Carrier


Must maximize

average backscattered

(carrier) power per

bit P0:

P0 σ1 + σ2 (scalar)

σi: Radar Cross Section

(RCS) at load ZL=Zi



Bit ‘0’ ( )Z1ZL= Bit ‘1’ ( )Z2ZL=

Carrier


Must minimize

variance of

Backscattered (carrier)

power Var(P0),

due to different bits

for M-bit message:

Var(P0) [(σ1 - σ2)2]/M

(scalar)

M≥96 for Gen2. We (can) drop this constraint!



1G

2G

As

O (0,0)

Bit ‘1’ ( )Z2ZL=

Bit ‘0’ ( )Z1ZL=

Must minimize Reader bit-error

probability (BER):


Design Example for Given Tag Antenna:

Passive Tag

• Solutions I and II provide the same error detection probability (Constraint 3).

• Solution II provides higher backscattered carrier power per bit (Constraint 1).


Tag Design

Experimentation with meander-type antennas, proposed in the literature for passive tags…

Battery-assisted tags: no need for power transfer => different problem…


Tag Design and Experimentation

Design…

Prototype…

Radar Cross Section can be increased compared to passive case (perfect matching).


Experimentation

Carrier transmitted…

Tag modulating waveform…

Received (backscattered) waveform…


Remarks

Tag design should maximize |Γ1-Γ2|

only…

Research is over in RFIDs…

Selection of best reflection

coefficients was given for any As (and

tag antenna, not necessarily minimum

scattering).

As can be easily computed in closed-

form, given measurements of scalar

RCS!

Method for closed-form

calculation of As was

omitted due to time constraints

BUSTE

DBUSTE

D


Current Focus

1. Read Range (frequency dependent)

2. Communication throughput (bps/sensor)

3. Scalability (number of RFID sensors) => anti-collision ability

4. Reading speed (number of sensors/sec): ~ 40 tags/sec current state of the art

5. Antenna size (as small as possible)

6. Packaging material & environment (affect reader/tag coupling)

7. Efficient tag manufacturing & programming

8. Tag/Reader Cost

9. Integration: addressing all (or most of) the above in an application!


Scalability of Backscatter Sensor Networks

For agricultural fields, sensor density is large (~1-1.5 sensor/m2)…

Large number N of sensors is needed…

Required bandwidth is proportional to Nδ…

Anti-collision performance depends on available bandwidth

…tradeoff between anticollision performance and N (or bandwidth)


Why not Zigbee (with 802.15.4)?

1. 5-10$ each in quantity of 1000) –

[however, the target is 1$ per node]

2. tx current ~30 mA @ 10 dBm,

rx current ~40 mA.

3. Speed 20kbps (868 Mhz), 40kbps (900+MHz),

up to 240 kbps (2.4GHz) – routing overhead?“ZigBee Architecture Overview”, available from ZigBee.org

1. cost of an MCU (1.5$(each) in quantity of 1!

[however, the target is 0.1$ per node with

ASIC]

2. tx current ~0.6 mA @ 1 MHz, no receiver

3. Speed a few bps – no routing overhead


Reader Antenna Capacity Enhancement

Beamforming antenna: Tag Collision occurs when tags close in modulating frequency AND close in geographical space…

=> Larger number of sensors for given bandwidth (compared to omni)!


BSN Capacity Enhancement with smart Reader Antenna…

Edge Collision probability is analytically derived as a function of

various uncertainties

fading uncertainty

modulation

location uncertainty

observation direction uncertainty

antenna directivity pattern


Anti-Collision with Reader Antennas…

modulations utilized in Gen2: inappropriate for high-density, extended range semi-passive tags…

… that is due to round trip nature of backscatter com…

quantified collision for any modulation, number of tags and given spectrum…

(useful for epc class 3 standard)

A. Bletsas, S. Siachalou, J.N. Sahalos, "Anti-collision Tags for Backscatter Sensor Networks", 38th European Microwave Conference (EuMC), October 2008, Amsterdam, Netherlands.

Bletsas, J.N. Sahalos, "Antenna Enhancements for Backscatter Sensor Networks", COST Antenna Systems & Sensors for Information Technology Societies (ASSIST) Workshop, April 2008, Limassol, Cyprus.

A. Bletsas, S. Siachalou and J.N. Sahalos, "Anti-collision Backscatter Sensor Networks", submitted June 2008, IEEE Transactions on Wireless Communications (TWC).


Reader Antenna Design in Practice

1

2

3

E. Vaitsopoulos, A. Bletsas, J.N. Sahalos, "On the RFID Design with Passive Tags and a Butler Matrix Reader", 13th Biennial IEEE Conference on Electromagnetic Field Computation (CEFC), May 2008, Athens, Greece.

Butler Matrix Feeding Network (BFN)


6-month focus: RFID in HealthcareMotivation = Medical Errors + Electronic Inventory Control

Paper-based environments: medical errors approach 40%

In-hospital Medication errors: 44,000 deaths per year in the US, 700 deaths per year in Canada. (Institute of Medicine, National Academic Press, 1999)

Theft of equipment/supplies: $4,000 per hospital bed each year ($3.9 billion annually in the US)

Asset Tracking: One third of personnel time is wasted in searching. ~10% of inventory is lost annually.


RFID Reader Antenna Transmit Diversity

1 Reader Antenna

2 Reader Antennas with passive splitter (3-dB Tx power loss per antenna)

z slice, z-field z slice, z-field

x slice, z-fieldx slice, z-field


References G. Vannucci, A. Bletsas, D. Leigh, "Implementing Backscatter Radio for Wireless Sensor Networks", IEEE Personal Indoor Mobile Radio Communications Conference (PIMRC), September 2007, Athens, Greece, pp. 1-5.

G. Vannucci, A. Bletsas and D. Leigh, "A Software-Defined Radio System for Backscatter Sensor Networks", IEEE Transactions on Wireless Communications (TWC), Vol. 7, No. 6, pp. 2170-2179, June 2008.

E. Vaitsopoulos, A. Bletsas, J.N. Sahalos, "On the RFID Design with Passive Tags and a Butler Matrix Reader",13th Biennial IEEE Conference on Electromagnetic Field Computation (CEFC), May 2008, Athens, Greece.

A. Bletsas, J.N. Sahalos, "Antenna Enhancements for Backscatter Sensor Networks", COST Antenna Systems & Sensors for Information Technology Societies (ASSIST) Workshop, April 2008, Limassol, Cyprus.

A. Bletsas, S. Siachalou, J.N. Sahalos, "Anti-collision Tags for Backscatter Sensor Networks", 38th European Microwave Conference (EuMC), October 2008, Amsterdam, Netherlands.

A. Bletsas, S. Siachalou and J.N. Sahalos, "Anti-collision Backscatter Sensor Networks", IEEE Transactions on Wireless Communications (TWC), submitted June 2008.

A. Bletsas, A. G. Dimitriou, J. N. Sahalos, “Improving Backscatter Radio Tag Efficiency“, COST RF/Microwave Communication Subsystems for Emerging Wireless Technologies (RFCSET) Workshop, April 2009, Brno, Czech Republic.

A. Polycarpou, A.G. Dimitriou, A. Bletsas and J.N. Sahalos, "RFID in Healthcare", COST Antenna Systems & Sensors for Information Technology Societies (ASSIST) Workshop, May 2009, Valencia, Spain.


Outline

1. RFIDs

2. Emergency Radio




Motivation

~2001 @ MIT: Distributed phased arrays…Code name: “Marblehead Island problem” (note: Marblehead is north of Boston)


Problem Formulation (1)

2009: make it more interesting…. No CSI at the transmitters… No feedback from the destination… No carrier sync availability…

Very stringent requirements as in Emergency Radio


Intuition


Problem Formulation (2)


Alignment Probability Calculation


Alignment Probability Results


Steady-State Alignment Probability Phase Offset Independence


Steady-State Alignment Probability Clock Frequency Skew Independence


Transient Alignment Probability Clock Frequency Skew Dependence


…and finally: beamforming gain and delay

Example:fc=2.4 GHzR = 1 Mbps => Ts = 1 μsecTc > 100 μsec => u < 1.25 km/sec

@ φ0 = π/4 => Lbf = 8.6 dB => ~4 symbols out 100

Thus, effective rate: = 1 Mbps x 4/100 @ Lbf = 8.6 dB= 40 kbps @ Lbf = 8.6 dB!

8.6 dB in power is a factor of 7.24 (!!!)


Remark (1)

“Distributed Beam-forming REQUIRES Carrier Synchronization and/or feedback from the destination”.

We provided zero-feedback, zero-CSI distributed beam-forming, based on unsynchronized carriers!

The proposed scheme could complement rescue workers (emergency radio) or reachback communication in wireless networks (e.g. low-cost sensor networks).

BUSTE

D


Remark (2)

“Marblehead Island problem” “Marathi Island Problem”


References

A. Bletsas, A. Lippman and J.N. Sahalos, "Simple, Zero-Feedback, Distributed Beamforming with Unsynchronized Carriers", submitted April 2009, IEEE Journal on Selected Areas of Communication (JSAC), Special Issue on Simple Sensor Networking Solutions.

A. Bletsas, A. Lippman and J.N. Sahalos, "Simple, Zero-Feedback, Distributed Beamforming for Emergency Radio", submitted April 2009, IEEE ISWCS 2009, Tuscany, Italy.


Outline

1. RFIDs

2. Emergency Radio


4. Bibliometrics & the Hirsch Index

5. Education trends and integration


The problem

1. One source, K half-duplex relays, several destinations…

2. Quasi-static SLOW fading (no temporal diversity)…

3. Network (global) CSI at relays or destinations: NOT AVAILABLE…

4. Low-complexity protocol: reduced coordination overhead… (it’s a network problem)

5. Low-complexity receivers, cheap radios…(in-band multiple transmissions = noise)


Approach

…always exactly two (2) in-band transmissions:

a) Source-to-destination

b) “best” relay-to-DIFFERENT destination!

“best” relay: selected opportunistically, reactively, in a distributed manner at MAC (no global CSI anywhere in the network)

“best” relay: best for epoch n-1, interfering for next epoch n

t


Approach “best” relay: best b for epoch n-1,

interfering i for next epoch n.

b≠i due to half duplex radios…

Scheduling invariant…

“Reactive” opposed to “Proactive”: The latter is energy-efficient but needs additional CSI.

Efficient MAC for opportunistic selection: “A Simple Cooperative Diversity Method based on Network Path Selection”, IEEE JSAC, March 2006.

Optimality proof for DF Relays: “Cooperative Communications with Outage-Optimal Opportunistic Relaying”, IEEE TWC, September 2007.

t

Major difference with prior art: interference is allowed, NO network/superposition/dirty-paper coding (low complexity protocol/receivers)!


Analysis

1. discrete, narrow-band, constant total tx power,flat-fading model under Rayleigh Fading.

2. reactive, opportunistic selection:

3. performance dependent on previous Epochs! (non-memoryless) => fix this by using bounds!

Outage probability given specific

interfering relay


Analysis

Need to compute all relevant outage probs, conditioned to (any) interfering relay i.

Example 1:

Example 2 (…little more involved):

Signal of interest (SOI) and Interfering Signal (IS)are fully correlated!!!

Selection Receiver (SR): simplest receiver, other simple receivers are also possible!


Results

…(more than) acceptable performance with weak or no SD link, weak inter-relay links!


Results

Opportunistic Relaying provides cooperative diversity and engineers the required outage probability plateau!

…thus, more efficient use of spectrum, no need for scheduling (from source) delays!

Outage probability plateau!


Remarks

Notion of useful relays in IaOR is redefined: “relays with strong paths towards source and destinations but weak links with each other!”

Opportunistic, reactive relaying engineers the appropriate plateau to mitigate “interference” and reduce scheduling delays!

No need for fancy coding (superposition or dirty-paper)……but need for efficient opportunistic selection (examples exist, more research is required and welcomed! )


Discussion

“relays with strong paths towards source and destinations but weak links with each other!”

Approach 1: find such relays given existing relaying densities in urban environments!

“Existing Urban Environments provide wifi terminal densities on the order of 1000 nodes/km2” [Jones and Liu 2007].

Approach 2: use directive antennas AT THE RELAYS!

Low complexity switched-beam antenna designs are possible [Vaitsopoulos, Bletsas and Sahalos, IEEE CEFC 2008]


Classic relaying vs classic network coding vs 2-way

physical network coding Classic network coding

(classic) 2-way physical network coding (PNC)

[figure from Katti, Gollakota and Katabi, ACM Sigcomm 2007]

Classic relaying


Classic network coding vs 2-way physical network coding

Classic network coding

(classic) 2-way physical network coding (PNC)

[figure from Popovski and Yomo, IEEE ICC 2007]


2-way physical network coding = 2-source separation

J. Hamkins, "An analytic technique to separate cochannel FM signals"IEEE Transactions on Communications, vol. 48, no. 4, April 2000, p. 543-546.

[figure from Katti, Gollakota and Katabi, ACM Sigcomm 2007]


Our contribution: 3-way Physical Network Coding

Generalize the Hamkins Algorithm to the case of 3 sources…

3-way PNC becomes possible…

Hamkins states in his paper that “such generalization is easy”.

Aggelos states that such generalization is not easy...

try it…

Must read: [Hamkins 2000], [Zhang, Liew, Lam 2006], [Katti, Gollakota, Katabi 2007]

][niCe


Remarks (2)

“Interference is veeeeeryyyy baaaaaaaaddddd”….

“We cannot live in interference”…BUSTE

D


References

A. Bletsas, A.G. Dimitriou and J.N. Sahalos, "Interference-Aware Opportunistic Relaying with Reactive Spectrum Sensing", submitted August 2008, IEEE Transactions on Wireless Communications (TWC).

A. Bletsas, J.N. Sahalos, "Interference-Aware Opportunistic Relaying with Reactive Spectrum Sensing", invited paper, IEEE Personal Indoor Mobile Radio Communications Conference (PIMRC), September 2008, Cannes, France.

A. Bletsas, A.G. Dimitriou, J.N. Sahalos, "Reduced-Delay Interference-Aware Opportunistic Relaying", to appear, IEEE International Conference on Communications (ICC) 2009, Dresden, Germany.

A. Bletsas, J.N. Sahalos, “3-way Physical Network Coding”, in preparation.


Outline

1. RFIDs

2. Emergency Radio



emerging radio …and the world

Documents

zituc telecom seminars

greecetuc telecom seminars

dimitrioutuc telecom

general tag antenna

tag design experimentation

passive tag solutions

optimal tag loads z1

antenna specificas