Energy Algorithm on NetSphere Simulation Result

1
Energy Algorithmon NetSphere Simulation Result

• Paskorn c.
• 6/16/05
• dssg_at_umb
• Version 2 (please ignore version 1)

2
Condition For Model 1 - 2

• These results are from the NetSphere simulator
  after implementing energy exchange algorithm to
  agent (CE).
• Simulation Cycle 1 second
• Run simulation for 9 Hours (simulation time)
• Window Size for Energy Exchange Algorithm 50

3
Condition For Model 1 - 2

• Only 1 platform (platform is not a biology
  entity)
• Only 1 Agent at first. Agent is implemented
  energy exchange algorithm.
• Agent
• initial energy 1000 energy unit
• replication threshold 3000 energy unit
• replication cost 100 energy unit

4
Model 1 Increase User REQ rate

• Input
• User Request Rate

• Each request give 10 energy units to agent

5
Result from simulation
The acquire energy is collected every 60
simulation cycle (60 sec). Thus, the energy
amount is 600 (60110) at time 0-2 hour, 1800
(60310) at time 2-4 hour, 3000 (60510) at
time 4-6, 4200 (60710) at time 6-8, and 5400 (
60910 ) at time 8-9. We can see the energy
line 0 along the graph because the simulator
collects the energy data before and after
simulation period. At the begin of each
simulation period, every 60 simulation cycle, the
acquired energy 0.
6
Result Number of agents
Number of Agent is increased
7
Model 2 User REQ changes
Input
8
Number of agents is changed
9
Model 3 Condition

• Platform is implemented Energy Exchange Algorithm
• Agent
• migrate when energy gt 1000
• Pay 100 for migration cost
• Migrate follow the user request source
• Replicate when energy gt 3000
• Pay 100 for replication cost
• Energy is divided to its child (half-half)
• Platform can migrate and replicate
• Replicate when
• Energy gt 5000
• found the healthier host (use healthy level) (not
  follow the user request source)
• Then eplication cost 100
• Dead when
• Energy lt 0
• No agent in this platform

10
Model 3 User REQ changes
Input (the same as Model 2)
11
Model 3 Number of Agent
12
Model 4 Number of Platform
13
Compare number of agent and number of Platform
14
Platform and Agents move towards to user request
source
START NODE
User Request Source
15
Result platform works on user request source
16
Next Step

• Calculate healthy level
• Homogenuous network
• Service adaptation
• The number of agents (availablity)
• Response time
• Single and multiple user locations
• Heterogeneous net
• Resourse adaptataion
• Assume ?
• Platform uses 30 healthy unit
• Agent uses 2 healthy unit

17
Paper GridNets

• Adaptive resource allocation on grid network by
  symbiotic architecture
• Introduction
• Design principles in the Netsphere architecture
• Agent
• Platform
• Energy Algorithm
• Symbiotic
• Simulation
• Related work

18
Simulation

• Show
• Energy exchange algorithm for 1 agent
• Maybe compare with EWMA or SMA
• Static network, heterogeneous hosts
• of agents in one platform (model 1)
• of agents and of platform
• Location of platforms
• resource utilization