Analysis of the effects of mobility on the grid location service in ad hoc networks

A mobile ad-hoc network (MANET) is a collection of wireless nodes that cooperatively form a network in the absence of any infrastructure or administration. The grid location service (GLS) is a well-known distributed location service that tracks mobile node locations in ad hoc networks. GLS can be used as a location service in a geographic-based routing protocol to considerably improve the scaling properties of mobile networks. Previous studies of GLS have used the random way point (RWP) mobility model for the performance evaluation. While RWP is useful, it may not capture some important mobility characteristics, of scenarios where MANETs may be deployed. In this paper we re-visit the GLS architecture and evaluate its performance over a wide, variety of parameter settings and mobility models. We study the behavior of GLS under different mobility models including Freeway, Manhattan and Group mobility models. These models capture interesting mobility characteristics like spatial dependence, temporal dependence and geographic restrictions. Our results show that protocol performance is indeed sensitive to the mobility model. We attribute this significant variance of performance to the interaction between the dynamics of the network connectivity and the protocol mechanisms, and analyze such interaction in our study. One interesting finding is that for group mobility, where; node movement is highly correlated and location discovery is conceivably simpler, the performance of GLS in fact degrades instead of improving. This different behavior for group mobility points to a subtle interaction and deeper coupling between the mobility model and the underlying grid.