Research On Techniques Of Pervasive Computing Oriented Semantic Reasoning Storage System

The computing model has been evolving toward the human centralized pervasive computing; accordingly the applications produce new behavior and requirements. As the fundamental software, operating system must also change to serve the applications better. On the other hand, many new research results have been gained to support this change. These results impel the operation system to improve itself despite the changes of computing model.This thesis researches on data storage mechanism in operating system. Analysis of pervasive computing by using RSS model (Rich Servants Service Model) concludes that intelligent stroge service is needed by pervasive applications, and the traditional file system can't satisfy this requirement due to the absence of data semantic information. To solve this problem, we focus our research on formal expression and reasoning ability of data semantic information and adding them to the storage system.As the latest research results of knowledge representation, the descriptive logic, SI is adopted to express the semantic information and provide the reasoning capability. When implementing the reasoning mechnisim, this thesis analyzes the problem with existing Block strategy in allusion to role inversion and proposes the Loop Block strategy. SI Tableaux algorithm is given based on it and proved to be the decision algorithm of SI. Inspired by the Copy-on-Write strategy, we give the COW Reasoning Tree to improve the memory usage and implement Backjump optimization on it. To quickly find collision condition during reasoning, we present a new optimization technology, Collision Concept Set Buffer. The incremental algorithm of SI reasoning is proposed to deal with the frequent additions and deletions of concept. The experiment shows that these technologies effectively enhance the reasoning performance, and the enhanced reasoning engine is capable of dealing the reasoning tasks in semantic storage system.We investigate the high-performance closed sequential mining algorithm, Bide. By analyzing Bide algorithm, the theorem of Backward-extension Event (BEE) detection is proposed and proved. It shows that BEE set of any prefix item is non-increasing with the extension of the prefix. And the accumulation performance of BEE set is optimized. The FCP tree is defined to represent the final result of FCP mining and its three characteristics are demonstrated. When the frequent item and the prefix are not coexistent in the new input sequence, the results of contiguous FCP mining are reusable. And the corresponding theorem is proved. The Bidelnc algorithm is proposed to incrementally mine FCPs. The experiment shows BideInc improves the mining performance in dynamic input sequence database.We can utilize it in the data prefetching mechanism to mine the block relationship.With the support of the above researches, we give the Sematic Reasoning Storage System (SRSS), including its runtime architecture, data view and physical architecture. The architecture of the Kernel Description Logic Engine (KDLE) is also presented.The conclusion can be drawn from the whole research that it is feasible to store semantic information and perform reasoning in operating system. Comprehensive requirements of pervasive applications can be satisified by this solution. It is demonstrated by our work that operating system will evolve considerably in pervasive computing era.