Cooperative Caching on Mobile Devices for Location Dependent Queries

Location dependent services are expected to become a very important revenue stream in the future. To enhance performance of location dependent queries there have been several research efforts in the past. Most of these have been focused on client side caching with extremely little work on cooperative caching. Cooperative caching might particularly be a very good solution to improve performance of such queries because they exhibit high degree of spatial locality. In this project we studied the potential of cooperative caching with some hands on experience. We compared an approach based on grouping and placement with naïve cooperative caching and a non-cooperative caching approach. This document describes the design, creation, and usage of a simulation system to explore the viability of cooperative caching. We proposed a new simulator design using simple events to efficiently implement a mobile environment with multiple clients and modified group mobility model. This simulator is highly configurable with the ability to add new maps and generically represent realistic approximations of mobility models like random waypoint model, reference velocity model, column group mobility model etc. Results show that cooperative caching is very promising and shows considerable improvement over non-cooperative caching. There are certain cases where cooperative caching with grouping and placement performs better than naïve cooperative caching. However, in most general cases cooperative caching with grouping and placement does not seem to offer much improvement over the naïve cooperative scheme. The gain might not be considered huge enough to motivate one to use such an approach in a commercial system.