Research On Cache Replacement Algorithm In Data Broadcast Environment

Recent advances in wireless data networking and portable computing, devices have engendered a new paradigm of computing, called mobile computing. In a mobile computing system, users access information anytime anywhere with their handled devices, even when they are moving. However, mobility constrains raises difficulties for mobile users, because mobile clients may often be disconnected from stationary server machines or may have a low bandwidth channel for sending messages to servers. At the same time, the characteristic of nodes storage resource limited in the embedded system affects the extensive popularity of application of the embedded system. Cache replacement algorithms is one of the important technology to conquer the limitation .Consequently, the study of the algorithm in data broadcast has great theoretical meaning and practical value.The frequent shift of user attention results the simple Cache manage algorithm (eg.LRU) can not satisfy the need of mobility environment. The paper studies mainly the data replacement algorithms based on cost which considers data access probability and broadcast frequency at the time. This paper introduces two cost-based replacement algorithms, including replace algorithm based on accessing histories and FIFO PIX replacement algorithm with lock. The former is the algorithm which forecasts the future visiting behavior according to the visiting history of user. Firstly the probability figure is fabricated by visiting log, and a data set with lesser visiting probability is screened out. Finally the replacement term with the biggest broadcast probability is selected. The latter is suitable for FIFO PIX with lock to avoid replacement of important data. Compared with traditional LRU algorithms, the two algorithms improve hit ratio and response delay time. The first way indicates that the improvement of hit ratio performance is especially obvious. The second way is simpler and reduces complex hardware circuit. The results of experiments indicate that they are practical algorithms, especially for handled devices with small capacity of storage.