Resource Coordination in Wireless Sensor Networks by Combinatorial - - PowerPoint PPT Presentation

Resource Coordination in Wireless Sensor Networks by Combinatorial Auction Based Method

Presented By:

• Md. Muhidul Islam Khan

Overview

§ Description of the problem § Challenges and design issues § Goals § Proposed approach § Simulation results § Conclusion and future works

Resource coordination in WSN

§ Dynamically assignment of available resources of sensor nodes. § Collaborative environment for resource management. § Allows sensor nodes to self schedule its tasks. § Helps to find out the best allocation of tasks to resources. § Helps to increase the lifetime of the network.

Resource Coordination Framework

Available ¡ Resource ¡ ¡ Infrastructure ¡ Coordina5on ¡Methods ¡ Goals ¡ Trigger ¡

Offline ¡ Periodic ¡ On-­‑demand ¡ Issued ¡by ¡changes ¡ in ¡network ¡ Resource Assignme nt Monit

• r

Challenges and Design Issues

§ What coordination method to use. § How to monitor a resource assignment. § How to model the WSN and application. § How to learn the best coordination.

Goals

§ Extend the lifetime of the network. § Communication with neighbor nodes by local information. § Tasks allocation and runtime adaptation of resources. § To learn the best scheduling of tasks.

Combinatorial Auction

§ Combinatorial auctions are a great way to represent and solve distributed allocation problems. § It is helpful when the bid is composed of several items. § Problem: most of the winner determination solutions that exists are centralized.

PAUSE Auction

§ The PAUSE (the Progressive Adaptive User Selection Environment ) auction is a combinatorial auction and the problem of winner determination is distributed amongst the bidders. § Provides a bidding algorithm for agents in a PAUSE auction, the PAUSEBID algorithm. § This algorithm always return the bid which is maximizing the bidders utility.

PAUSE Auction

§ The most widely used auction in multi-agent systems. § Agents can place bids for sets of items. § Has been used in many systems where there is a set of tasks need to distributed between agents with different preferences.

PAUSE Auction

§ Combinatorial Auction is applicable to a large number of distributed allocation problems and multi-agent coordination problems § PAUSE auction has been developed and distribute the winner determination problem among agents § The existing centralized auctions don’t fit multi-agent systems where: § Agents have own computational resources § Agents have localized information

The PAUSE Auction

§ Each bid b composed of bitems :the set of items the bid is over bvalue : the value or price of the bid bagent : the agent that placed the bid b {bitems , bvalue , bagent} § At the end of each stage k, set B { b1, b2,b3,..} of the current best bids is generating and all agents know the best bid for every subset

• f size k or less.

The PAUSE Auction

At each stage: § Bidders can use bids from other agents from previous round. § The sum of bid prices in each submitted bid set should be bigger than currently winning bid set. § There will be a set of currently winning bids which maximizing the revenue.

The PAUSE Auction

§ Also at each stage the goal of each agent i is to maximizing it’s utility where vi is the value function for this agent is: § Agent i must find g* such that

Bidding Algorithm

Bidding Algorithm

Bidding Algorithm

j j j

D S j Objective γ η β α + + = . ) (

j

S

j

η

j

D

α β

γ

Signal Strength of Node J. Resource price. The distance between the target and the node. Equilibrium Constants.

Bidding Algorithm

Ideal Time Gap

IG

T

CompCost refers to Computation cost E refers to remaining energy

) )( ( ) , (

2

f R e I V CV f R CompCost

CPU nV V dd dd CPU

T dd

+ =

) ( c V K f

dd −

≅

Simulation

Required CPU cycle: Sensing: 10 M Hz Transmit: 26 M Hz Receive: 26 M Hz Sleeping: 0 Energy requirement: Sensing: 0.0000841 J Transmit: 0.00233 J Receive: 0.00231 J Sleeping: 0.000012 J Ideal time gap: Sensing: 25 msec Transmit: 10 msec Receive: 10 msec Sleeping: 0

Simulation

Simulation

Object Tracking Application Using Our Approach

Node A Node B Node C

Tasks Executions Over Time Steps

Node A Node B Node C

1 ¡ 7 ¡ 13 ¡ 19 ¡ 25 ¡ 31 ¡ 37 ¡ 43 ¡ 49 ¡ 55 ¡ 61 ¡ 67 ¡ 73 ¡ 79 ¡ 85 ¡ 91 ¡ 97 ¡ Time ¡Steps ¡

Tasks ¡Execu0ons ¡for ¡Node ¡A ¡

Sleep ¡ Sense ¡ Transmit ¡ Receive ¡ 1 ¡ 9 ¡ 17 ¡ 25 ¡ 33 ¡ 41 ¡ 49 ¡ 57 ¡ 65 ¡ 73 ¡ 81 ¡ 89 ¡ 97 ¡ Time ¡Steps ¡

Tasks ¡Execu0ons ¡for ¡Node ¡B ¡

Sleep ¡ Sense ¡ Transmit ¡ Receive ¡ 1 ¡ 9 ¡ 17 ¡ 25 ¡ 33 ¡ 41 ¡ 49 ¡ 57 ¡ 65 ¡ 73 ¡ 81 ¡ 89 ¡ 97 ¡ Time ¡Steps ¡

Tasks ¡Execu0ons ¡for ¡Node ¡C ¡

Sleep ¡ Sense ¡ Transmit ¡ Receive ¡

Conclusion and Future Works

§ Our method helps to increase the lifetime of the network. § Simulation results show better performace. § To design our state space with more system variables. § To consider some other tasks like aggregation in our system. § To apply other learning algorithms and to find out their efficiency. § To find out the scalability of this approach.

THANK YOU !!!! muhidulislam.khan@a au.at