Research on scheduling algorithms adapting to arrival rate changes of stream data

在本研究项目中，我们研究了动态到达率流数据的调度算法，并调查了围绕流数据处理的研究趋势。为了处理以不可预测的速率连续到达的流数据，调度算法必须动态地适应到达率的变化。在本研究项目中，开发了基于二进制流数据的共享窗口连接的动态调度算法。与以往的连续查询调度不同，我们的算法不仅考虑了查询的吞吐量，而且考虑了每个查询的处理速率，同时适应了流数据到达率的变化。实验结果表明，我们的算法能够以较小的调度成本实现良好的调度。在对流数据处理研究趋势的研究中，我们在安全和隐私、P2P系统、SMT处理器处理和web缓存系统等领域取得了一些成果。实际上，我们已经讨论了隐私保护计数聚合查询到k分类，基于规则的细粒度数据库访问控制完整性检查，使用副本实现负载平衡的P2P协议，SMT处理器考虑程序阶段的线程调度方法，以及在P2P网络中引入组框架来构建web缓存系统。我们还明确了这方面的几个问题，供今后的工作参考。

In this research project, we have studied on scheduling algorithms for stream data with dynamic arrival rates and have also investigated research trends around stream data processingFor handling stream data which arrive continuously at unpredictable rates, scheduling algorithms have to adapt to the changes of arrival rates dynamically. In this research project, dynamic scheduling algorithms for shared window joins over binary stream data have been developed. Unlike previous work on scheduling of continuous queries, our algorithms take account of not only the throughput of queries but also the process rate of each query, while they adapt to the changes of arrival rates of stream data Some experimental results show that our implementation of the algorithms can work well with small scheduling cost.In studies on research trends around stream data processing, we have some results concerning such fields as security and privacy, P2P systems, processing in SMT processors, and web caching systems. In fact, we have discussed privacy preserving count aggregate queries to k-classification, rule-based integrity checks for fine-grained database access control, a P2P protocol for load balancing by using replicas, a thread scheduling method for SMT processor considering program phase, and introducing group framework into P2P networks for constructing web caching systems. We have also clarified several problems in such fields for future work.