An asymmetric cost model for query processing in mobile computing environments

We address the issue of query processing for information services in mobile environments, using a mobile sales and inventory application example. Efficient query processing depends upon choosing one of several candidate query plans based upon a cost model which accurately reflects computation and communication costs. Unlike traditional distributed database environments, mobile environments necessitate new cost models which are asymmetric in the sense that the resources available at mobile sites are less than those at fixed sites, and communication costs vary with the direction of data transfer. In addition, communication costs may vary as the mobile moves. We design a cost model which incorporates these criteria and illustrate it via examples. As costs change dynamically, it is desirable that the query execution plan be modified dynamically to optimize the cost. Existing distributed query processing algorithms can be modified to suit the mobile environment and this paper discusses our ongoing work to do so.