【作者】 洪亮；

【导师】 卢炎生；

【作者基本信息】 华中科技大学 ， 计算机软件与理论， 2009， 博士

【摘要】 计算技术和无线通讯技术的发展与结合使得一种全新的计算模式——移动计算模式成为了现实。采用了移动计算模式的计算机网络被称为移动计算环境。移动计算环境下最重要的数据服务之一是位置相关数据服务，它是指计算机网络向网络用户提供数据查询的结果，该结果依赖于查询指向的移动对象／用户的位置。移动计算环境下的位置相关数据服务的研究尚处于起步阶段，并日益成为一个热点。其中在无线蜂窝网络中，如何利用缓存的方法提高定位移动用户的性能；在资源受限的无线传感器网络中，如何处理基于事件的位置相关查询，并有效管理传感器的移动性；在移动无线传感器网络中，如何高效地连续处理位置相关聚集查询等是研究的重要方面，具有重要的理论与实际意义。无线蜂窝网络中位置相关的数据服务的一个关键研究点是移动对象/用户的位置管理，其中提高定位移动用户性能的一个重要方法是缓存移动用户的位置信息。然而已经提出的缓存策略针对的是单个用户，造成缓存效率不高。针对群体用户，一种基于位置数据库聚类的动态适应缓存位置信息(简称DACaL)策略提高了移动计算环境下的位置管理的性能。其中位置数据库聚类算法通过挖掘群体移动用户的运动模式对位置数据库进行聚类，以确定缓存层次和降低位置管理的代价。动态适应缓存位置信息算法根据聚类结果对位置数据库进行重组，在相邻聚类之间缓存位置信息，建立旁路指针，以缩短消息传输的路径和减少查询位置数据库的次数。已有的研究工作很少涉及无线传感器网络中的位置相关数据服务，因为在动态的、分布的和资源受限的网络中对移动结点进行管理和定位，并连续处理查询是一个难点。一种基于事件的位置相关数据服务查询(Event-based Location DependentQuery，简称ELDQ)模型可以持续地聚集用户感兴趣的移动传感器周围一定区域内的传感器数据。事实上，ELDQ模型概括了许多典型的数据服务查询并且存在于大量应用中。当前的查询处理方法不能够高效地处理ELDQ。因此，有必要研究相应算法和技术来处理ELDQ同时优化系统性能，包括自适应代理选择算法、网络内查询分发算法、网络内查询传播算法、基于位置的网络内聚集算法和两级多查询优化算法。由于传感器的移动性，移动无线传感器网络中的位置相关查询要求持续更新查询结果。然而已提出的方法需要固定基础设施的支持并且没有利用连续快照查询(Snap-shot Query)之间的联系。移动无线传感器网络中一种不需要固定基础设施的位置相关查询处理方法优化了处理连续位置相关查询(Continuous LocationDependent Query，简称CLDQ)的总体代价。该方法包括一种基于跳数阈值的分发路径更新算法、一种基于冲突的距离感知消息调度算法和一种基于位置预测的连续查询处理性能优化方法。更多还原

【Abstract】 The development and combination of computing technologies and wireless communication techinologies have made a new computing mode—mobile computing mode a reality. The computer networks which use mobile computing mode are called mobile computing environments. Location dependent data service is one of the most important data sercices in mobile computing environment. Such data service provides query results that are dependent on the mobile objects/users’ location to the network users. The research on location dependent data services in mobile computing environment is still at an initial stage and increasingly becomes a hot spot. The dissertation addresses three research issues: improving the performance of locating mobile users using the caching method in wireless cellular networks; processing event-basesd location dependent data queries and efficiently managing the sernsors’ mobility in resource-constrained Wireless Sensor Networks (WSNs); efficiently processing continuous location dependent aggregation queries in mobile WSNs.Location management of the mobile objects/users is one of the key research topics of location dependent data services in wireless cellular network. To improve the performance of locating mobile users, one important approach is caching mobile users’ location information. However, existing caching strategies only consider a single user, which is inefficient. We propose a Dynamic Adaptive Caching Location (DACaL) strategy by location database clustering for mass mobile users. DACaL strategy reduces the total cost of location management in mobile computing environment. In LDB-Clustering, location databases are clustered by mining mobile users’ moving pattern to determine the caching level and reduce location management cost. In Dynamic-Adaptive-Caching, location databases are reorganized based on clustering result, location information is cached and bypass pointers are created between adjacent clusters to shorten signal traveling path and reduce times of querying location databases.Few existing research work has addressed the location dependent data services in wireless sensor networks. The reason is that it is difficult to manage and locate mobile sensor nodes in dynamic, distributed and resource constrained netoworks to continuously process the query. We propose an Event-based Location Dependent Query (ELDQ) model that continuously aggregates data in specific areas around mobile sensors of interests is presented. In fact, ELDQs generalize several typical data service queries, which are important in many applications. However, existing approaches fail to efficiently answering ELDQs. Therefore it is necessary to research corresponding algorithms and techniques including adaptive proxy selection, in-network query dissemination and propagation, Location-based In-network Aggregation (LIA) and two-level multi-query optimization to process ELDQs while optimizing system performance.Due to sensor mobility, location dependent queries in mobile WSNs require continuous query results. However, existing approaches require fixed infrastructure and do not exploit the relationships between successive snap-shot queries. We propose an infrastructure-free approach for processing location dependent queries in mobile WSNs to optimize the total processing cost of Continuous Location Dependent Query (CLDQ). This approach consists of a Hop-based Dissemination route Update (HDU) algorithm, a Contention-based Distance-aware Message Scheduling (CDMS) algorithm and a performance optimzaition approach for continous query processing based on location predictions.更多还原