Energy-Conserving Access Protocols for Identification NetworksBy Imrich Chlamtac, Chiara Petrioli, and Jason RediIEEE/ACM TRANSACTIONS ON NETWORKING, Feb. 1999

CSE519 Embedded Networks

Feb. 5, 2007

Su Jin Kim

Overviews

Introduction Current Access Protocols Proposed Protocols Energy-Analysis Simulation Results Conclusion References

Introduction

Radio Frequency Identification Devices (RFID) and Infrared Identification Devices (IRID) Small, Inexpensive, resource-limited

IDNET (IDentification NETwork) Interconnected base stations Large number of small low-cost wireless tags Tags contain microprocessor power source, a RF receiv

er, transmitter, and some support logic. -> Active Tags

RFID Systems

Examples: Location tracking of the animals, Supply chain, Health-Care etc.

Characteristics Scale: large Cost: inexpensive Size: small Traffic: short, simple message

→ Important issues : Low Energy and Low Delay Requirements

Current Access Protocols (1)

Low Power Design awake & sleep state

Random Access Protocol (Aloha) The base stations send packets at random times The tags awakes at random times The probability of a tag and the base station being

awake in the same slot is very low High the energy consumption and packet delay

Current Access Protocols (2)

Classical TDMA Assign a time slot to each tag Low energy requirement

awake 1/N slots, N = # of tags in the system High packet delay

Trade-off: the energy vs. delay How frequently does a tag awake?

Network Model

N tags share a radio channel Packet-oriented and packet length is constant The time is slotted and the base station’s transmission is

synchronized Exactly one packet can be transmitting during each slot Access Protocol

Transmission Scheduling: at the base station The base station selects a packet for transmission from the arrival

queue in each slot Wake-schedule: at each tag

The tag determines the slots being awake

Grouped-Tag TDMA Protocols

Divide tags into m = N = # of tags in the system X = # of tags in the group

Assign each slot to one group Increase the average energy consumption per slot Decrease the average delay Drawbacks

Tags continue to wake up cyclically The packets’ destination distribution is heavily clustered, the

performance can degrade severely

xN /

Directory Protocols

The base station1. waits for k packets2. Broadcast the directory which lists the destinations of the k

packets3. Transmit the actual packets

Tags 1. listen to the directory and find out when they wake up2. When there is no group being transmitted, the tags wake up

periodically every v slots Trade-off

Small k, v: Low Delay, but High energy consumption Large k, v: High Delay, but Low energy consumption

Pseudorandom Protocols

All tags1. run the same pseudorandom number generator, but each tag has

the unique seed

2. Determine their state (awake or sleep) based on a probability p

3. Stored state of the random number generator

The base station Know the seeds of tags Possible to determine the schedules of tags Change p based on tags’ expected traffic rates

Good for the heterogeneous traffic patterns

Energy Analysis (1)

E: average percentage of slots in which a tag is awake Grouped-Tag TDMA Protocols

E =

m = # of the groups in the system

=

m

1

xN /

Energy Analysis (2)

Directory Protocols E =

k = # of packets in the group

k’ = the slots need for transmitting the directory

Pseudorandom Protocol E = p

'

'

kkNk

k

Simulation Results

15,000 packets N = 1000 tags Inter-arrival rate,

I = 0.05, 0.2, 0.5 arrivals per slot

Classical Access Protocols

* Random Access

Only when p is high (> I), the system is stable

* Classical TDMA

Good Energy Consumption (0.001)

Extremely Long Delay

(≥500 slots)

Grouped-Tag TDMA with uniform destination distribution

•X = large

High Energy Consumption

Low Delay

•X = small

Low Energy Consumption

High Delay

• FINDING the OPTIMAL X is IMPORTANT!

Directory Protocol with uniform destination distribution

•k = large

Low Energy Consumption

High Delay

•With given k, v = large

Low Energy Consumption

High Delay

• FINDING the OPTIMAL k, v is IMPORTANT!

Pseudorandom Protocolwith uniform destination distribution

• Slightly worse than the grouped-tag TDMA

• But, the difference decreases as I increases

Energy Conserving Protocols with I = 0.2

Energy Conserving Protocolswith wide Gaussian Destination Distribution

• The performance of the grouped-tag TDMA degrades rapidly

•The performance of the pseudorandom degrades very slightly

Energy Conserving Protocolswith narrow Gaussian Destination Distribution

• With I = 0.5, the performance of the grouped-tag TDMA is completely unstable

Conclusion

Classical TDMA and Random (such as Aloha) Access Protocols are not appropriate for the RFID Systems

Uniform Distribution

Moderate Distribution

Heavy Distribution

Low traffic load

Grouped-Tag TDMA

Pseudorandom Pseudorandom

High traffic load

Pseudorandom Pseudorandom Directory

References

“Energy-Conserving Access Protocols for Identification Networks,” I. Chlamtac, C. Petrioli, and J. Redi, IEEE/ACM Transactions on Networking, Vol. 7, No. 1, Feb. 1999

“Analysis of Energy-Conserving Access Protocols for Wireless Identification Networks,” I. Chlamtac, C. Petrioli and J. Redi, the Proc. of Int. Conf. on Telecommunication System, March 20-23, 1997

“Extensions to the pseudo-random class of energy-conserving access protocols,” I. Chlamtac, C. Petrioli and J. Redi, the Proc. 2nd IEEE Int. Workshop Wireless Factory Communication Systems, Oct. 1997, pp. 11-16