Data management in wireless environment
Weili Wu, Jiaofei Zhong · 2012
Wireless Data Broadcasting is a new developed data dissemination method for spreading public information to a mass number of mobile subscribers. Access Latency and Tuning Time are two main criteria to evaluate the performance of a data broadcasting system. With the help of indexing technology, clients can reduce tuning time significantly by searching target data items through indices and turning to doze mode while waiting. Many different indexing schemes are developed under data broadcasting environment, such as distributed index, Hash scheme, etc. In this dissertation, several most popular indexing schemes for data broadcasting systems were redesigned, including distributed index, Huffman tree index, Hash scheme, and exponential index. A unified communication model was created, a novel evaluation strategy by probability theory was constructed to formulate the performance of each scheme theoretically, and the simulations were conducted to compare their performance by numerical experiments. Our communication model can easily be modified by service providers to satisfy their specific requirements. They can also evaluate other indexing schemes using our model, in order to choose the best scheme for their systems. Sensor database system is another popular and well-explored database system with sensor nodes as data sources that not only collect data but also combine stored data to provide information for the users who issue queries. Queries will be disseminated across the sensor database system, where each sensor generates data from the area it covers, and transmits the related data to the original sensor. Due to battery limitation of sensors, it is crucial to select a minimum yet sufficient subset of sensors to cover the query region while maintaining connectivity, in order to diminish the communication cost and energy consumption. This problem is defined as the Sensor Database Coverage (SDC) problem. In this dissertation, a polynomial-time approximation algorithm for the sensor database coverage problem was introduced. It is shown that when the number of targets is much less than the number of sensors, the algorithm has a performance ratio of O(log 3 n), which is better than existing algorithms. Experiments were conducted to evaluate the performance, and the results confirm the efficiency of our algorithm.