Modeling and analyzing the impact of location inconsistencies on geographic routing in wireless networks

Abstract

Recently, geographic routing in wireless networks has gained attention due to several advantages of location information. Location information eliminates the necessity to set up and maintain explicit routes, which reduces communication overhead and routing table size. These advantages allow scalability especially in dynamic and unstable wireless networks. However, no matter which technologies or techniques a location system uses, its measurements will have some amount of quantifiable inaccuracy depending on environment and system. These inaccuracies may affect the performance and even correctness of geographic routing. However, thus far, these impacts have not been studied in-depth. In this paper, we analyze the impact of location inaccuracy on geographic routing. First, we model location inaccuracy metrics - absolute location inaccuracy, relative distance inaccuracy, absolute location inconsistency and relative distance inconsistency. Then, we analyze how location inaccuracy metrics affect the building blocks of geographic routing including greedy forwarding and local maximum resolution. We use extensive NS-2 simulations to evaluate the performance of geographic routing using a wide array of parameter settings including inaccuracy level, node degree and network diameter. In our simulation results, reasonable location inaccuracy (of 20 percent or less of radio range) caused packet drop reaching up to 54 percent, non-optimal path up to 53 percent and packet looping. These observations indicate the importance of re-visiting geographic routing protocols, and the significance of considering location inaccuracy in their design and evaluation.