NetLogo User Community Models

(back to the NetLogo User Community Models)

Download If clicking does not initiate a download, try right clicking or control clicking and choosing "Save" or "Download".(The run link is disabled for this model because it was made in a version prior to NetLogo 6.0, which NetLogo Web requires.)

## WHAT IS IT?

Routing is a critical issue in the Delay Tolerant Network as finding the most optimized path, with minimum loss of data and in the shortest span of time is a challenge. Various
researchers have proposed different strategies for detection of the effective and efficient path.In this model we apply a metaheuristic method of ant colony optimization (ACO) to propose a routing mechanism for Delay Tolerant Networks to find the
optimized path with maximum throughput. Modifications have been carried out in ACO to incorporate characteristics of another meta-heuristic technique known as particle swarm
optimization for building up effective routing scheme. Experiments on simulated networks using agent-based modeling tool named NetLogo shows that our technique performed better
as compared to the original ACO. Here the throughput has been specially taken as the performance factor of the network.
This algorithm addresses two distinct problems. First, given an arbitrary collection of nodes which are constantly changing their position and subsequently transmission links with other nodes, the routing algorithm converge quickly to shortest
paths which consequentially changes over time. Thus, algorithm will adjust quickly to reflect changes to the network. Secondly, transmitting packets along the same path may result in undesirable congestion in network and over utilization of a given set of nodes used for transmission, which preferably should be avoided.

## HOW IT WORKS

Ant colony Optimization (ACO) has a chiefly studied drawback of converging into local optima and the proposed enhancement is intended at overcoming this prevailing disadvantage in ACO. Numerous research in the past has been devoted at achieving so . In the real scenario, ants travel vast areas without global view of the ground. No central processor exists to lead the ACO towards good paths. Convergence is certain, but time to converge is indeterminate. Taking primarily this shortcoming in account, we have tried to integrate particle swarm optimization characteristics to improve the traditional ACO. Previously global information has been employed in the form of global pheromone updating rule, to place or deposit extra pheromone to predispose the exploration procedure from a non-local viewpoint . Our modification to ACO is motivated by both study into diverse forms of ant colonies systems and their distinction with other meta-heuristic practices. In particle swarm optimization, each particle denotes a contestant solution and these particles are moved around in the search-space according to simple computed formulations over location and velocity of the particle. Each particle's movement is not subjective towards its local best known position but, also inclined towards the best discovered positions in the search-space, which are improved as better positions are located by other particles. This is expected to move the swarm toward the best solutions. Here we define a probability global-opt-factor, which correlates to the chance that a packet will follow best path discovered so far but only if the last best set path has not been deformed to change in nodes position and present node of the packet exists in the best-path.(further explained in the algorithm). In DTNs, the routing technique must adapt to topological modifications (e.g., link/node failures, link/node addition/removal). Therefore the convergence of the algorithm must be rapid and shortest routing paths may sustainably change over time. Our modification speeds up the convergence and accounts for global best as an update to each ant’s solution depending upon the global-opt-factor.

## HOW TO USE IT

The "num-of-nodes" slider determines the numberof the nodes in the network.
"num-packets" initializes the model with the number of packets.
The transmission range of nde is determined by the "Transmission Range" slider.
"Simulation Time" slider specifies the timesteps to run the model for.
"Minimum Node Speed" and "Maximum Node Speed" denote the limits of the nodes' speed.
SETUP intializes the model with nodes positions according to Hybrid Topology.
GO-FOREVER runs the model with the algorithm governing the transmission of packets.

## THINGS TO NOTICE

When we set the congestion factor to a low value=0.5, we observe that
packets follow the same path between the source and destination. On the contrary, a high congestion factor packets will distribute throughout the network.
Check the throughput of the algorithm compared to the traditional ant colony algorithm.

## THINGS TO TRY
Genetic Algorithm:
Use the following code in R to determing optimal parameters of the algorithm

library(GA)
library(RNetLogo)
path.to.NetLogo <- "C:/Program Files (x86)/NetLogo 5.0.4" #Netlogo path here
NLStart(path.to.NetLogo, nl.version=5)
model.path <- "<model path here>"
NLLoadModel(model.path)

f <-function(cf,gf){
NLCommand("setup")
NLCommand("set congestion-factor",cf)
NLCommand("set global-opt-factor",gf)
NLCommand("operate")
x=NLReport("total-packets")
y=NLReport("total-generated-packets")
ratio = x/y
ratio
}
GA <- ga(type = "real-valued", fitness = function(x) f(x[1],x[2]), min = c(0,0), max =c(6,1),popsize = 25)

## CREDITS AND REFERENCES

Bhatia, Abhishek, and Rahul Johari. "Genetically optimized ACO inspired PSO algorithm for DTNs." Reliability, Infocom Technologies and Optimization (ICRITO)(Trends and Future Directions), 2014 3rd International Conference on. IEEE, 2014.
Special thanks to: xa.yimg.com/kq/groups/4872827/1000694762/name/lab4.pdf