+ All Categories
Home > Documents > Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

Date post: 03-Jan-2016
Category:
Upload: caldwell-mullins
View: 27 times
Download: 1 times
Share this document with a friend
Description:
Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks. Zhipeng Cai , Shouling Ji , Jing (Selena) He, Anu G. Bourgeois Georgia State University. OUTLINE. 1. Introduction. System Model. 2. Distributed Data Collection. 3. 4. Simulation and Analysis. 5. - PowerPoint PPT Presentation
Popular Tags:
28
Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks Zhipeng Cai, Shouling Ji, Jing (Selena) He, Anu G. Bourgeois Georgia State University
Transcript
Page 1: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

Zhipeng Cai, Shouling Ji, Jing (Selena) He, Anu G. BourgeoisGeorgia State University

Page 2: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

2

OUTLINE

3

Introduction1

2

4

System Model

Distributed Data Collection

Simulation and Analysis

5 Conclusion

Page 3: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

3

Introduction

Page 4: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

4

Cognitive Radio Networks (CRNs) The utilization of spectrum assigned

to licensed users varies from 15% to 85% temporally and geographically (FCC report)

Unlicensed users (Secondary Users, SUs) can sense and learn the communication environment, and opportunistically access the spectrum without causing any unacceptable interference to licensed users (Primary Users, PUs)

Introduction

Page 5: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

5

Why Distributed Algorithms? CRNs tend to be large-scale distributed systems

CRNs are dynamic Systems

Spectrum opportunities are dynamic with respect to time and space

Challenges How to guarantee secondary network activities do not hurt primary

network activities?

How to make decision based on only local information?

How to overcome problems induced by lack of time synchronization?

How to theoretically analyze the performance of distributed algorithms?

Introduction

Page 6: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

6

Contributions Derive a Proper Carrier-sensing Range (PCR) under the physical

interference model for Secondary Users (SUs)

Propose an order-optimal Asynchronous Distributed Data Collection (ADDC) algorithm

Simulations are conducted to validate ADDC

Introduction

Page 7: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

7

System Model

Page 8: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

8

Primary Network N independent and identically distributed (i.i.d.) PUs

Locally finite property

Working power

Network time is slotted with slot length

During each time slot, each PU transmits data with probability

System Model

Page 9: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

9

Secondary Network n SUs and one base station

Maximum transmission radius of SUs is r

The secondary network can be represented by graph

Conditions on communication between two SUs

System Model

Page 10: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

10

Data Collection At a particular time slot t, every SU produces a data packet of size B

The set of all the n data packets produced by SUs at time t is called a snapshot

The task of gathering all the n data packets of a snapshot to the base station without any data aggregation is called a data collection task

The data collection delay is the time consumption to finish a data collection task

The data collection capacity is the average data receiving rate at the base station during a data collection process

System Model

Page 11: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

11

Interference Model Physical interference model

For PUs

For SUs

System Model

Page 12: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

12

Distributed Data Collection

Page 13: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

13

Data Collection Tree Proper Carrier-sensing Range (PCR) Data Collection Algorithm Performance Analysis

Distributed Data Collection

Page 14: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

14

Connected Dominating Set (CDS) based Data Collection Tree

Data Collection Tree

Page 15: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

15

Objectives The secondary network does not cause unacceptable interference to the

activities of the primary network

All the SUs transmitting data simultaneously are interference-free

The carrier-sensing range is as small as possible, which implies SUs can obtain more spectrum opportunities

Proper Carrier-sensing Range

Page 16: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

16

Concurrent Set: a set of active nodes s.t. all the nodes in this set can conduct data transmission simultaneously.

:

Proper Carrier-sensing Range (PCR): the carrier-sensing range R is a PCR if for any R-set, it is a concurrent set.

Proper Carrier-sensing Range

si

Page 17: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

17

How to decide the proper carrier-sensing range (PCR)?

In a R-Set, to guarantee SUs will not cause unacceptable interference to PUs, it is sufficient to have

(Lemma 2)

In a R-Set, to guarantee SUs can transmit data simultaneously and interference-freely, it is sufficient to have

(Lemma 3)

We can set the PCR , where

Proper Carrier-sensing Range

Page 18: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

18

Proper Carrier-sensing Range

Page 19: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

19

Asynchronous Distributed Data Collection (ADDC) algorithm

Data Collection Algorithm

Page 20: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

20

The number of dominators and connectors within the PCR of an SU is upper bounded by , where is a function on x with

(Lemma 5)

The number of SUs within the PCR of an SU is upper bounded by

, and with probability 1.(Lemma 6)

The expected time for an SU to obtain a spectrum opportunity is

where . (Lemma 7)

Any SU having data for transmission can transmit at least one data packet to its parent within time .

(Theorem 1)

Performance Analysis

Page 21: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

21

The delay induced delay by the proposed Asynchronous Distributed Data Collection (ADDA) algorithm is upper bounded by

This implies the achievable data collection capacity of ADDC is

which is order-optimal. (Theorem 2)

Performance Analysis

Page 22: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

22

Simulation

Page 23: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

23

Network setting An i.i.d. primary network

An i.i.d secondary network

Please refer to the paper for detailed settings

Compared algorithm Coolest (ICDCS 2011): the path with the most balanced and/or the

lowest spectrum utilization by PUs is preferred for data transmission

Simulation

Page 24: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

24

Data Collection Delay vs. Network Size (n and N)

Simulation

Page 25: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

25

Data Collection Delay vs. and

Simulation

Page 26: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

26

Data Collection Delay vs. Transmission Power

Simulation

Page 27: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

27

We study the distributed data collection problem in CRNs

We propose an Asynchronous Distributed Data Collection (ADDC) algorithm for CRNs, which is order-optimal

Simulations are conducted to validate the performance of ADDC

Conclusion

Page 28: Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

THANK YOU!

Optimal Distributed Data Collection for Asynchronous Cognitive Radio Networks

Zhipeng Cai, Shouling Ji, Jing (Selena) He, Anu G. BourgeoisGeorgia State University


Recommended