|My B.Tech thesis (completed as of May'08) is titled 'Effect of Power Control on Forwarding strategies in Wireless Ad-hoc networks'. The starting point was a paper titled 'Transmission range control in multihop packet radio networks' by T.C. Hou and V.O.K Li. The paper compared the relative performance of the major forwarding strategies - NFP (Nearest with most Forward Progress), MFR (Most forward with Fixed Radius) and MVR (Most forward with Variable Radius). These strategies are properly described in the paper by Li and also in my thesis and presentations (see below for the links). The comparison was on parameters like one-hop throughput, interference at one-hop receiver, overall transmission progress etc.
My modification in oder to stud these strategies was to include the Interference Zone in the analysis. Other nodes within this circular zone of a particular node can interfere with their own transmissions, hence increasing the probability of decoding errors. Firstly, a non-power control case was handled i.e. transmissions with full power for all nodes. The limits for the number of potential interferers within the Interference zone was established and then a probabilistic relation was assigned to the interference caused by a certain fixed number of nodes in the zone (within the limits). This was a complex triple integral over the interference zone. Since the expression was not closed-form, a computer program was made to get the expected interference values, providing with some insightful results.
Thereafter the equations were modified for a general power-control case. The order of the integral increased, hence the time-complexity of the program also went up. With a change in the value of a single parameter (a probability distribution function modified from the paper by Li), the results could be obtained for all the strategies. The parameters evaluated were single-hop interference and SINR (at receiver) and power utilized at the transmitter.
What was surprising was that a random forwarding strategies i.e. selection of intended receivers at random was found to be more-or-less at par with the other strategies w.r.t the evaluated parameters. As in the paper with Li, NFP was found to be the best strategy, except for low node densities for which either of NFP or MFP/MVR proved to be equally good.
Please see the files below to get and in-depth account.
Click here to view final thesis
Click here to view final presentation
Click here to view mid-semester presentation