Online Data Gathering for Maximizing Network Lifetime in Sensor Networks

29
Online Data Gathering for Maximizing Network Lifetime in Sensor Networks IEEE transactions on Mobile Computing Weifa Liang, YuZhen Liu

description

Online Data Gathering for Maximizing Network Lifetime in Sensor Networks. IEEE transactions on Mobile Computing Weifa Liang, YuZhen Liu. Outline. Introduction System model Generic Cost Model of Energy Consumption Problem Definition Algorithms For Online Data Gathering - PowerPoint PPT Presentation

Transcript of Online Data Gathering for Maximizing Network Lifetime in Sensor Networks

Page 1: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

Online Data Gathering for Maximizing Network Lifetime in Sensor Networks

IEEE transactions on Mobile Computing

Weifa Liang, YuZhen Liu

Page 2: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

2

Outline

Introduction System model Generic Cost Model of Energy Consumption Problem Definition Algorithms For Online Data Gathering Performance Evaluation Conclusions

Page 3: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

3

Introduction

The main constraint of sensor nodes is their low battery energies, which limit the networks lifetime.

The network lifetime of a wireless sensor network is defined as the time of the first node failure in the network.

Energy efficiency in the design of routing protocols for sensor network is of paramount importance.

Page 4: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

4

Introduction

To prolong the network lifetime, there are many different energy optimization metrics have been proposed.– Minimize the total energy consumption– Maximize the lifetime of each node

There are two typical forms of data gathering queries:– Periodic collection.– Event driven.

Page 5: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

5

Introduction

Consider an online data gathering problem in sensor networks:– There is a sequence of data gathering queries,

which arrive one by one.– To respond each query as it arrives, the system

build a routing tree for it.– Within the tree, the volume of the data transmitted

by each internal node depends on not only the volume of sensed data by the node itself, but also the volume of data received from it children.

Page 6: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

6

Introduction

The objective is to maximize the network lifetime without any knowledge of future query arrivals and generation rates.

The experimental results show that, among the proposed algorithms, one algorithm that takes into account both residual energy and the volume of data at each sensor outperform the others.

Page 7: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

7

System Model

Consider a wireless sensor network consisting of :– N stationary sensor nodes and a base station S distributed

over a region.– Location of each sensor and base station are fixed and

known a priori.– Every sensor equipped with an omni-directional antenna.

We take into account the transmission energy consumption only and assume that the other energy consumption such as reception are negligible.

Page 8: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

8

System Model

The wireless sensor network can be modeled by a directed graph M = (N,A).– N is the set of nodes, A is the set of edges,– Edge <u,v> in A if node v is within the transmissio

n range of u.– For u, v with distance , the transmission energ

y at node u is modeled to be proportional to if a unit of message is transferred from u to v.

– is a path-loss exponent parameter depending on the characteristics of communication medium.

,u vd

,u vd

Page 9: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

9

Generic Cost Model of Energy Consumption – Symbol Definition

Given a data gathering query, we aim to build a routing tree T rooted at the sink node.

For each node v:– p(v) be the parent of v in T.– be the length of the sensed message by v– A relay node v in T , assume the lengths of messages that v

received from t children are– be the amount of energy consumption of receiving an

d sensing a unit of message by a sensor.

vm

1 2, ,..., tl l l

,e sr r

Page 10: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

10

Generic Cost Model of Energy Consumption

v may or may not aggregate these messages and its own sensed message before transmitting them as a single message to its parent p(v).

The length of message transmitted by v to its parent is a function f with parameters and

1 2, ,..., tl l l vm

Page 11: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

11

Generic Cost Model of Energy Consumption

The definition of f may vary, depending on application domains. There are two frequently used f below:– The length of message transmitted by a relay

node is independent of message lengths of its children and itself.

– The length of the message transmitted by a relay node depends on the message lengths of its children and itself.

Page 12: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

12

Generic Cost Model of Energy Consumption

If v is a relay node with childrenthe cost c(v) of v is thus–

If v is a leaf node, f is a function of only the cost c(v) of v is thus–

1 2, ,..., tu u u

1 2 , ( )1

( ) ( , ,..., , ) * *t

t v v p v i e v si

c v f l l l m d l r m r

vm

, ( )( ) ( )( )v v p v sc v f m d r

Page 13: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

13

Generic Cost Model of Energy Consumption-Three well-known model

Case1: Assume that– Total transmission energy consumption is

considered:– Minimum residual energy among the nodes is

considered:

1 2( , ,..., , )t vf l l l m k

1 2 , ( ) , ( )( ) ( , ,... , )t t

t v v p v v p vv V v V

C T f l l l m d k d

, ( )( ) min{ ( ) }t

v p vv V

C T re v kd

Page 14: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

14

Generic Cost Model of Energy Consumption-Three well-known model

Case2: Assume that– Total transmission energy consumption is considered:

– Minimum residual energy among the nodes is considered:

Case3 is the mixture of case1 and case2 that takes into account both optimization objectives simultaneously.

1 2( , ,..., , ) ( )t tf l l l m kD v

1 2 , ( ) , ( )( ) ( , ,..., , ) ( )t t

t v v p v v p vv V v V

C T f l l l m d k D v d

, ( )( ) min{ ( ) ( ) }v p vC T re v kD v d

Page 15: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

15

Problem Definition

The online data gathering problems is to maximize the network lifetime without knowledge of future query arrivals and the generation rate.

As a query arrives, the response by the system to query is to build a routing tree rooted at the sink and spanning the other nodes for it.

All in all, the problem is to maximize the number of queries answered until the first node in the network fails.

Page 16: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

16

Algorithms for Online Data Gathering-MNL

The proposed algorithm is that, once a data gathering query arrives, a data gathering tree for the query is constructed using a greedy policy that maximize the minimum residual energy among the nodes.

If node v is included into the tree if it leads to maximizing the minimum residual energy.

The nodes are included into the tree one by one.

Page 17: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

17

Algorithms for Online Data Gathering-MNL Symbol Definition

T and : the tree and the set of nodes included in T.

Initially, the set of nodes in T contains the sink node only.

Each time the algorithms picks up a node v from such that is maximized.

The algorithm continues until all nodes are checked

TV

TV Vmin{ '( ) | { } { }, , }Tre u u V v s v u E

Page 18: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

18

Algorithms for Online Data Gathering-MNL

Let node be the considered node.1. If there are l edges from v to the nodes in

denoted by , wherefor all I , define

2. Otherwise( there is not any edge from v to the nodes in ), define

Tv V V

TV

1 2, , , ,..., , lv u v u v u i Tu V1 i l

, , ( ) ,( , ) min{ ( ) , ( ) | , }i ii v u u p u u sg v u re v kd re u kd u P u s

max 0( ) ( , ) max{ ( , ) |1 }l ig v g v u g v u i l

TV max 0g

Page 19: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

19

Algorithms for Online Data Gathering-MNL

Page 20: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

20

Algorithms for Online Data Gathering-MNL

Page 21: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

21

Algorithms for Online Data Gathering-MMRE

Maximization of the minimum residual energy (MMRE) among the nodes in the network.

An inverted spanning tree instead of a broadcast tree is constructed.

Let T be tree tree and be the set of nodes in T. The sink node s in included in T andinitially.

Each time it picks up a node if v satisfies

The Algorithm continues until

TV{ }TV s

Tv V V

', '', '

'( ) max { ( ') | ' , ' }v u T Tv u E

re v re v kd v V V u V

0TV V

Page 22: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

22

Algorithms for Online Data Gathering-SPT

Minimization of the total transmission energy consumption of relaying the sensed message from a sensor to the sink node.

An energy graph G(V,E) is derived from the sensor network.– V: the set of sensor nodes and sink node s– Directed edge <u,v> in E from u to v if the residual energy at u is at

least – The weight assigned to the edge is , which is the energy consu

mption of transmitting a init message between of two nodes A single-source shortest path tree rooted at the sink node is co

nstructed. The minimum transmission energy consumption to send its k-u

nits sensed message to the sink node is Thus, the total transmission energy consumption is

,( )v skW P

,u vkd

,u vd

,( )v sv V

kW P

Page 23: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

23

Algorithms for Online Data Gathering-BT

An undirected, energy graph for the sensor network is defined.

– V: set of sensor nodes – E: set of undirected links

We aim to prolong the network lifetime by dealing with two opposite optimization.

– Minimize the total energy consumption of all nodes– Minimize the total energy consumption by each node

However, constructing a spanning tree that meets these two optimization objective is NP-hard

( , , )G V E

Page 24: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

Balancing Minimum Spanning and Shortest Path Trees, 1993

24

Algorithms for Online Data Gathering-BT

An approximate algorithm that balances these two optimization objectives is available, and a solution delivered by the proposed algorithm is within times of the optimum [15].

Despite algorithm BT taking the total energy consumption for realizing a data gathering query into consideration, it doesn’t take into account the residual energy at each individual node.

This results in the nodes near the tree root running out of their batteries quickly.

( ', ')

Page 25: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

Efficient Algorithms for Finding minimum spanning Trees in Undirected and Directed Graphs

25

Algorithms for Online Data Gathering-MDST

Is similar to the one of algorithm BT, but different weight function is used.

The weight assigned to a link is

Where Beta is the energy utilization ratio at node u, between its consumed energy and its initial capacity.

,u v E ( )

,( , ) ( 1)uu vu v d

Page 26: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

26

Performance Evaluation

Page 27: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

27

Performance Evaluation

Page 28: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

28

Performance Evaluation

Page 29: Online Data Gathering for  Maximizing Network Lifetime in Sensor Networks

29

Conclusions

The experimental results showed that algorithm MNL significantly outperforms the others algorithms including MDST, MMRE, SPT, and BT.

An Algorithm take the total energy consumption for realizing a data gathering query into consideration, it would result in the nodes near the tree root running out of their batteries quickly after a number of data gathering queries are realized, since those nodes always relay messages for the other nodes.