CS 712 | Fall 2007

Using Mobile Relays to Prolong the Lifetime of Wireless Sensor NetworksWei Wang, Vikram Srinivasan, Kee-Chaing Chua.National University of Singapore.

MobiCom’05

27th September ‘07

JunSup LeeDependable Software LAB. KAIST

1

CS 712 | Fall 2007

Agenda

Problem Statement Motivation Overview Related work Contents Experiment & Conclusion

2

CS 712 | Fall 2007

Problem statement

WSN in inaccessible and hostile environments Nodes are randomly scattered Nodes are expected to self-organize to form a multi-hop network

Sink : gateway to a backbone network Relay : relaying information from other sensors to the sink

Existing approaches to improving WSN lifetime Device research

Design low power hardware and efficient energy sources Network research

Design efficient algorithms at the various networking layers

3

Limitation : They assumes entire network is composed of static nodesLimitation : They assumes entire network is composed of static nodes

CS 712 | Fall 2007

Motivation

A large sensor field composed of Static nodes : simple, cheap, large numbers Mobile nodes : rich, expensive, a few

Question Should a mobile node act as a sink or a relay? What is the lifetime of the network for each of these alternatives? What are the trade-offs associated with each of these alternatives?

4

Only static nodes

bottleneckbottleneck

CS 712 | Fall 2007

Overview

Investigate the benefits of a heterogeneous architecture for WSN composed of Rich mobile nodes (Small #)

Mobile relays Mobile sinks

Static nodes (Lange #)

Mobile nodes act as Relays or Sinks ? Investigate the performance(lifetime) in different routing algorithm Case 1 : when the network is all static Case 2 : when there is one mobile sink Case 3 : when there is one mobile relay

5

Because : distributed the bottleneck nodes all around the networkBecause : distributed the bottleneck nodes all around the network

Using mobile node as a sink results in the maximum improvementUsing mobile node as a sink results in the maximum improvement

CS 712 | Fall 2007

Overview

However, in hostile terrains, it is not always possible for the sink to be mobile Mobile nodes as Relays (and gives a joint mobility and routing algorithm)

To ensure performance, mobile nodes need to stay only within a two hop radius of the sink It requires all the nodes to be aware of the location of the mobile node.

Proposed Algorithm Mobile nodes as Relays (and gives a joint mobility and routing algorithm)

Achieves same performance as the mobile sink Requires only a limited number of nodes to be aware of location of the mobile

Only nodes within a certain distance of the sink need to be aware of the location of the mobile relay

Experiment Compare the performance of the mobile relay and mobile sink

For a densely deployed sensor field of radius R hops Proposed algorithm requires O(R ) mobile relays to achieve the same performance as the

mobile sink

6

CS 712 | Fall 2007

Related Work

Mobile nodes as mobile sinks to save energy Randomly moving Predictable and controllable moving

Disadvantage Delays : Static nodes need to wait for the sink to approach it Sink monitoring : Static nodes have to be aware of the location of the mobile

sink to route information Maintain connection : It will be difficult to engineer a system whereby a

mobile sink is connected at all times to the backbone network

Approaches in this paper Static sink Mobile nodes act as relays

In a large and dense network Mobile nodes only need to patrol in a two hop radius around the sink

7

CS 712 | Fall 2007

Comparison of Energy Conserving Methods

Mobile relay Static routing

The mobile relay decides a schedule, which id given by a sequence

li : location of node i ui : duration of time the mobile relay spends at that location

Dynamic routing Sensors need to know where the mobile relays are and redirect their packets

according to the position of the mobile relays The mobile can attract more traffic to pass through it and the traffic through

bottle neck nodes can be decreased.

8

CS 712 | Fall 2007

Mobile Relays In Dense Network | Assumption and Network Model

Assumptions Sensors are densely deployed

Poisson point process Density : λ Circular area : Radius R >> 1

Dense? The packet can travel as far as the transmission range in any direction

Data logging application Sensors send their sensing data at certain rate Date generation rate, Transmission range is same

Energy consumptions Energy of the sink is unlimited Sensor nodes have the same initial energy : E Total energy consumed by sensors in sending out one packet : e Average number of packets flowing out from the sensor k per time unit : fk

9

Sink n0

N nodes

R >> 1

CS 712 | Fall 2007

Mobile Relays In Dense Network | Upper Bounds on Lifetime

Static sensor set According to distance to the sink

Set of nodes which can reach the sink with minimal hop count i : Pi

Pk =

Set of nodes which can reach the sink within j hops

Qj =

Upper bound Lifetime of static network : Lifetime of dense network with one mobile node (upper bound)

The least upper bound on lifetime is when i = 2

Mobile Relay needs to stay only within a two hop radius of the sink in order to maximize the lifetime

10

CS 712 | Fall 2007

Mobile Relays In Dense Network | Joint Mobility and Routing Algorithm

Aggregation Routing Algorithm (ARA) Joint mobility and routing algorithm

In a dense static network with one mobile relay Life time is close to the upper bound The mobile relay needs to only stay within a two hop radius

Aggregation node in P3 : #a3 > #a1 + #a2; ; Process

P1 : Directly send to the sink P2 : Send to none aggregation nodes in P3 (P3/G) P3 : Send to P3/G , P3/G relay data to aggregation nodes Else : Relay the packets to line OM, OM nodes relay until the packets reach the

aggregation point

11

CS 712 | Fall 2007

Mobile Relays In Dense Network | Routing With Few Nodes Aware of Mobile Location

Aggregation Routing Algorithm with Limited Nodes (ARALN) ARA problem

All nodes need to know mobile node location

ARALN Only limited nodes have to aware the location of mobile node : inside nodes of Qs Aggregation Ring : After packet reach Ps

Process : Switch(k:the index of Pk where n is element of Pk) Case S+1, ……. , R :

Shortest path to the sink Case S :

Send to current Aggregation ring (k-1+r) or next ring (K-2+r) Case 4, ….. S-1 :

Send to OM line until it travel in Pk or reach to OMand send to next ring (K-2+r)

Case 1,2,3 Call method ARA;

Lifetime

12

CS 712 | Fall 2007

Mobile Sink VS Mobile Relays | Performance Compare

The lifetime improvement of two mobile approach Mobile Sink

times better than the all static network Mobile Relays

With R mobile nodes, we can achieve improvement as the mobile sink solution

Network with m mobile relays M mobile nodes in the network will stay within and get nearly 4m times

lifetime when R is large

13

CS 712 | Fall 2007

Experiment Results & Conclusion | Network Lifetime

Static network compare with one mobile node 200 nodes and moderate node density Simulation is based on simplified energy model Static network uses linear optimization algorithm (because of this R<= 4)

14

CS 712 | Fall 2007

Experiment Results & Conclusion | Lifetime Improvement with One Mobile Relay

Simulation result

15

CS 712 | Fall 2007

Q&A

Thank you ^^

More information? jslee@dependable.kaist.ac.kr

16