Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile...

21
Mobile Agent Migration Problem Yingyue Xu

Transcript of Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile...

Page 1: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Mobile Agent Migration Problem

Yingyue Xu

Page 2: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Energy efficiency requirement of sensor networks

• Mobile agent computing paradigm• Data fusion, distributed processing in sensor

networks

Motivation

Page 3: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Client/Server-based Computing

Several clients One server Clients send data to server Server processes these data

Drawbacks: Creates heavy network traffic and consumes

bandwidth, resulting in poor performance Dependent on the performance of the server

Page 4: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

A New Paradigm: The Mobile-agent-based Computing

Mobile agent is a special kind of software

Migrate from node to node, carrying partially integrated results and performing data processing autonomously

Data stay at the local site, while the processing task is moved to the data sites.

Page 5: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Itinerary– Route of migration

Identification– Unique for each mobile agent

Data buffer– Carries the partially integrated results

Method– Execution code carried with the agent

Architecture of Mobile Agent

160.10.30.100

itinerarydata buffer

method

identification

Page 6: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Problem Definition

• The problem of finding an optimal sequence of nodes for a mobile agent to visit in order to complete its task in the minimum expected costs.

Page 7: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Background: Travel salesman problem

Page 8: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Travel salesman problem

o given a finite number of "cities" along with the cost of travel between each pair of them, find the cheapest way of visiting all the cities only one time and returning to your starting point.

o NP-complete: computational effort required to solve this problem increases exponentially with the problem size.

o Single mobile agent, full connection, single visit problem

Page 9: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Background

• Vehicle routing problem• to deliver a set of customers with known demands on

minimum-cost vehicle routes originating and terminating at a depot.

• NP-complete• Special case is the Travel Salesman Problem• multiple mobile agents, full connection, single visit

problem

Page 10: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Exact Approaches: propose to compute every possible solution until one of the bests in reached

o Branch and bound

• Heuristics: perform a relatively limited exploration of the search space and typically produce good quality solutions within modest computing times

o Constructive methods

• MetaHeuristics: the emphasis is on performing a deep exploration of the most promising regions of the solution space.

o Genetic algorithms

o Ant algorithms

o Tabu search

Solution Techniques

Page 11: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Single or multiple mobile agents• Single or multiple visits• May not be full connection

Mobile agent migration

Page 12: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Static methods: centralized methods, compute the route in advance of mobile agent migration.

o Can get good solutions

o Energy inefficient!

o Time consuming

• Dynamic methods: determine the route locally, on the fly.

o Suitable for sensor networks

o Need to find a way that minimize searching cost, while

achieving satisfying solutions

Mobile agent migration problem

Page 13: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Katsuhiro Moizumi, “Mobile agent planning problem”, Dartmouth College

o Information retrieval

o Conclusion: sorting the ratios for the sites,

(ti+l)/ pi, into increasing order and visiting the sites in that order

Related work

Page 14: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Qishi Wu, “On computing the route of a mobile agent for data fusion in a distributed sensor network”

o Maximizing an objective function, shown to be NP hard

o Using genetic algorithm to solve optimization problem

o Develop a simulator in VC to simulate

o Used in our simulations

Related work

Page 15: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Once the mobile agent arrives a node, it randomly selects a destination from its neighbors

• Easier to implement

Random selection algorithm

Page 16: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Greedy algorithm

• Once the mobile agent arrives a node, it selects a destination from its neighbors that

Page 17: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Using Java• Implement random selection, greedy

algorithms• For optimal path, we use the software from

LSU to calculate the path, then simulate in our simulator

Simulation

Page 18: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Parameters

• 30m by 30m • Grid deployment• Transmission range: 30m• Target position in the center of the area• Accuracy threshold: 0.9

Page 19: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Results

100 120 140 160 180 200 220 240 2600

0.002

0.004

0.006

0.008

0.01

0.012

0.014

0.016

Number of nodes

Tot

al e

nerg

y co

nsum

ptio

n

random selectiongreedy optimal

Page 20: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

Results

100 120 140 160 180 200 220 240 2600

0.002

0.004

0.006

0.008

0.01

0.012

0.014

0.016

Number of nodes

Tot

al e

nerg

y co

nsum

ptio

n

random selectiongreedy optimal

100 120 140 160 180 200 220 240 2600

5

10

15

20

25

30

35

40

45

50

Number of nodes

Num

ber

of h

ops

random selectiongreedy optimal

Page 21: Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.

• Greedy algorithm is the best in term of energy consumption and hop number

• The optimal algorithm may not always return the best solution

Conclusions