presented by zhu jinshun season: shelving interference and joint identification in large- scale rfid...

Presented by Zhu Jinshun

Season: Shelving Interference and Joint Identification in Large-scale RFID Systems

How RFID works

Typical RFID system

Basic Tag Assembly

I. Introduction

II. Preliminaries

III. Season

IV. Performance Evaluation

V. Related Works

VI. Conclusion

Outlines

Introduction

Collisions in RFID systems

Anti-collision methods

tag

reader

Collisions in RFID systems

Tag collision Reader collision Reader-Tag collision

Collisions in RFID systems

Tag collision Reader collision Reader-Tag collision

contentious region contentious tag

Anti-collision methods

Tag collision

Reader collision

Reader-Tag collision

FSA

Colorwave

Protocol stack: Season

I. Introduction

II. Preliminaries

III. Season

IV. Performance Evaluation

V. Related Works

VI. Conclusion

Outlines

Preliminaries

A. Tag Collision

B. Reader Collision

C. Reader-Tag Collision

D. System Model

Tag Collision

Mainly employ Time Dividing Multiple Accesses (TDMA)

A popular anti-tag-collision algorithm

is Framed Slotted ALOHA (FSA)

Tag collision

Slot

Slot

Slot

………………

frame frame ……procedure：

Tag Collision - FSA

Tag Collision-FSA

Command ：request, select, read/write, quit, reset

Types of slots ：Idle, single, collided

Reader Collision

Reader Conflict Graph (RCG)Colorwave

Reader collisionRCG

Reader - Tag Collision

schedule (r1,r2 ) in sequence, we can only consider the other two types of collisions

Reader-Tag collision

System Model

I. Introduction

II. Preliminaries

III. Season

IV. Performance Evaluation

V. Related Works

VI. Conclusion

Outlines

Season

A. Observations

B. Overview

C. Season-I

D. Season-II

E. Season-III

F. Discussion

Season - Observations

Majority of tags are non-contentiousContentious tags cause the major delayThe signals from contentious tags can be

received

Season - Overview

Season is a protocol stack two phases ： Shelving Interference （ Season-I ） Joint Identification （ Season-II ， Season-III ）

Season - Overview

Season-I ： non-contentious tagsSeason-II ： active reader ， passive readerSeason-III ： data from contentious tags

Season - Season-I

concurrently identify tags the majority of tags can be identifiedSimilar to FSA ， frames ， equivalent

slotsdifferent from FSA ， tune the length of

frames

Season - Season-I

One slot = 1/f the optimal choice of f isUSE ， estimate Dynamically adjust the frame to -k

after the k-th tag is collected

SeasonSeason - Season-II

Joint Identification has two advantages:

1 、 can avoid reader collisions among neighboring readers

2 、 Reduce the identification delay

significantly

Season - Season-II

Active reader and passive readerActive readers ,two conditions:

1 、 cover edges as most as possible

2 、 will not incur signal interference among themselves when concurrently activated

Season - Season-II

the Maximal Weighted Independent Set (MWIS)

Notation ： v ， CH(v) ， JOIN(v,u) ， EXIT

v ， either a clusterhead or an ordinary node

Season - Season-II

Season - Season-II

Season - Season-III

Cross-range tag collision

Cross-range tag collision, a new tag collisionSeason-III

to allow active and passive readers to identify contentious tags collaboratively

Season - Season-III

Given that the set of active readers is A and the set of passive readers is the P

On one hand, for active readers…On the other hand, for passive readers…

Season - Season-III

On one hand, for active readers:

estimate the number of contentious tags ， TDMA ， keep the tags in active state util all “FINISH”

broadcast “silence”

Season - Season-III

On one hand, for passive readers:

listen , estimate

send a “FINISH” message

if no neighboring passive readers,

otherwise, next scheduling round

Season - Season-III

For example

Season - Discussion

Unbalanced Loads of Readers

session numberSource Sensitive and Insensitive

I. Introduction

II. Preliminaries

III. Season

IV. Performance Evaluation

V. Related Works

VI. Conclusion

Outlines

Performance Evaluation

A. Evaluation Methodology

B. Implementation Results

C. Simulation Results

Evaluation Methodology

Testbed and deploymentSimulating Real RFID ApplicationsPerformance Metrics

Evaluation Methodology

Testbed and deploymentuse a NI PXI-1044 RFID testing tool with PXI 5600 receiver as our passive reader

interrogation range 2m

deploy five readers

find that the percentage of contentious tags is less than 10%

Testbed and deploymentSimulating Real RFID ApplicationsPerformance Metrics

Evaluation Methodology

Simulating Real RFID Applicationstwo typical application scenarios and three random reader topologies

Warehouse: 72 readers, 78,606 records Object tracking: 1653 records, includes the tag

locations, source, and identification time Random Topologies: “Sparse”,“Moderate”,“Dense”

Evaluation Methodology

Performance Metrics Throughput: the ratio of total number of tags to the overall

identification time Average Delay Read Rate: environment noise,multi-path, signal attenuation Scheduling Round: the efficiency of anti-reader-collision

Evaluation Methodology

Implementation Resultsemploy a NI PXI-1044 testing tool with a PXI 5600 receiver as the passive reader, Alien reader as the active reader

We can observe that the passive reader achieves a read rate of 0.73 in 60% of testing cases. The average value of its read rates is up to 0.71, which is nearly as good as that in the single-reader deploying scenario.

Identifying tags without reader collisions

when number of tag is above 100, Season-I has 30.6% and 42.2% time saving on average than BT and FSA

Identifying tags without reader collisions

Season-I, up to 0.4 and 60% of the cases has a throughput higher than 0.37

Identifying tags with reader collisions

Identifying tags with reader collisions

Identifying tags with reader collisions

the average delay of Season is no more than 300 time slots

I. Introduction

II. Preliminaries

III. Season

IV. Performance Evaluation

V. Related Works

VI. Conclusion

Outlines

Related Works

FSA, Colorwave, USE……

I. Introduction

II. Preliminaries

III. Season

IV. Performance Evaluation

V. Related Works

VI. Conclusion

Outlines

Related Works

Anti-collision is a crucial task

we propose an anti-collision protocol stack, Season

Our results show that Season significantly increases throughput, dramatically reduces the delay for tags

Thanks !