A data broadcast scheme based on prediction for the wireless environment

Broadcasting is an essential paradigm to transport data and support scalability in the wireless environment. Traditional index-based broadcast schemes take advantage of tree structures for saving energy but they record too much index information to make the broadcast cycle short. On the other hand, the hash scheme may suffer from longer broadcast cycles and data collision problems, resulting in longer access and tuning time. In this paper we propose a novel broadcast scheme, named predictive indexing, which allows mobile clients to access data by predicting data locations in the broadcast cycle. The new scheme not only has very compact structure, but also is flexible enough to allow the server to adjust the index structure for saving energy or reducing access time. Experimental studies show that it outperforms the tree-based schemes 2.5 times in energy saving, while maintaining roughly the same access time. It also outperforms the flexible indexing 70% and the hash scheme 80% in energy consumption.