Redes Manet

MANET extensions to ns2
Andr´s Lagar Cavilla e andreslc@cs.toronto.edu Department of Computer Science University of Toronto
This is a set of software modules providing additional models for the simulation of multihop mobile ad hoc networks (MANETs) in the ns2 simulator. We provide implementations of the CM and Shell mobility models, and the AF and LOS radio propagation models described in [1].We also provide instrumentation tools for the ns2 simulator, as well as template simulation scripts and results parsers. The software presented here builds on the Monarch ns2 wireless extensions [2], as well as on some pre-existing mobility and radio propagation models, such as Random WayPoint [3], Free Space and Two-Ray Ground. We assume that the reader is familiar with the concept of MANETs,with the models aforementioned, and with the internals of ns2. All the software is based on version 2.26 of the ns2 distribution. This documentation is split in two sections. The first part is a walk-through around the provided software, explaining how to conduct a simulation similar to the ones we carried out in [1]; we have bundled in the tarball the necessary scenario files. In the second part we godeeper into the fundamentals of each model and the particular implementation at hand. The software is provided “AS IS”, i.e. we do not take any responsibility whatsoever on any damage it may cause to your system or data. As long as you give proper credit, you can use the software freely for whatever legal purposes you may wish, and you can also modify it at will.

1
1.1

The Walk-ThroughMobility Pattern Generation

If you are reading this, then you have successfully decompressed the tarball, and are now wondering where to start among all those directories. The first thing you want to do is go to the setdest directory and build the four flavors of setdest we have there. You need to edit Makefile to have the NSDIR variable point to the root of your ns2 source tree. Once you’ve donethat, make should do its job with no problems. A typical simulation scenario in [1] was a network of 40 nodes moving at 3 m/s during 1200 s, following the CM mobility model, with no pause time in an area of 88 m by 113 m, and with a sensitivity threshold of -81 dBm assuming AF propagation. To randomly generate a mobility pattern according to those specifications, you have to type 1./setdest_cm_af -n 40 -r -81 -p 0 -s 3 -t 1200 -x 88 -y 113 -f bahen_cm_graph.txt -g bahen_af_pic.pnm -d bahen_af_defs.txt > mobility-40-cmaf-81.1 This will take a while, and the mobility pattern will be written to the mobility-40-cmaf-81.1 file. You may want to generate 4 more mobility pattern files with the same specifications, labeled mobility-40-cmaf-81.2 to mobility-40-cmaf-81.5. This way you can run 5different experiments and average the results. setdest does not only generate a mobility pattern, but also computes some topology metrics derived from that pattern, and appends them as comments at the end of the output file. The next step is then to go to the parsers directory and compile the Java program MobilityParser by typing javac MobilityParser.java. Then you can move the resulting .class file towhere the mobility pattern files are located, and run java MobilityParser mobility-40-cmaf-81 5. This will average the mobility metrics in each of the 5 files you have generated, and dump the results to mobility-40-cmaf-81.csv. This is a comma-separated-values file, readable by Matlab and any spreadsheet program. This csv file in particular will contain the values for three metrics: Link Changes: a globalcount of how many times the connectivity between each pair of nodes is interrupted and restored. Neighbor Density: the average number of peers within connectivity range for any given node at any given time. Optimal Path Length: the average length in hops of the optimal path between every pair of nodes for which a path actually exists, at any given time. Now you have five mobility pattern files...