Using reordering technique for mobile transaction management in broadcast environments

As computer hardware and wireless network technologies are developed to a high degree, there are many research efforts which intend to utilize data broadcasting to a large population of mobile clients through wireless channels. In recent years, different models of data delivery have been explored, particularly the periodic push model where the server repetitively disseminates information without any explicit request. In this paper, we devise new transaction processing algorithms, O-Post for update transactions and O-Pre for read-only transactions, in broadcast environments. Basically, each client executes its transactions in an optimistic manner and does some consistency checks based on periodic invalidation reports. When any kind of conflicts is found, the conflict order is determined according to the notion of reordering and the remaining operations are executed to hold the decision. We also develop a cache algorithm to cope with frequent restarts due to the optimistic execution. Experimental results are given to show the benefits of the proposed algorithms.