Ken C. K. Lee

Semantic query caching in a mobile environment

Caching of remote data in a mobile clients local storage can improve data access performance and data availability. Traditional approaches are page-based, without taking advantage of the semantics of cached data. It is difficult for a client to determine if a query could be answered entirely based on locally cached data, forcing it to contact the database server for additional data. We propose a semantic caching mechanism which allows data to be cached as a collection of possibly related blocks, each of which is the result of a previously evaluated query. We investigate mechanisms for transforming projection-selection queries to reuse cached data blocks. This avoids transmitting unwanted data items over low bandwidth wireless channels. Cache replacement techniques based on the semantics of cached data are also proposed. We describe the design of our prototype and study its performance.

An efficient algorithm for predictive continuous nearest neighbor query processing and result maintenance

Predictive continuous nearest neighbor queries are concerned with finding the nearest neighbor objects for some future time period according to the current object and query locations and their motion information. Existing continuous query processing algorithms are not efficient enough, requiring multiple dataset lookups to evaluate the query results throughout the duration of a continuous query. More importantly, the complete result for the whole query time interval is only available at the moment when all object motion updates have been examined, based on which adjustment of the query result is made. In this paper, we propose an algorithm which requires only one dataset lookup to deliver a complete predictive result. We then apply a differential update technique to maintain the query results incrementally in the presence of object location and motion updates.

Semantic data access in an asymmetric mobile environment

The mobile environment is inherently asymmetric. To utilize the downstream bandwidth effectively, hot data items should be disseminated over the broadcast channel to the mobile clients. To equip clients with the ability to identify the nature of the broadcast and to determine the answerability of their queries, semantic descriptions are associated with information units in the broadcast, organized into data chunks. Based on the nature of data items received over the broadcast, clients can initiate appropriate requests to make up for the remaining items over the back channels in an on-demand basis. We investigation into several semantic-based algorithms to organize data chunks and compare their performance with traditional data item-based organization in the asymmetric environment.

Incremental view maintenance for mobile databases

Abstract. In a mobile environment, querying a database at a stationary server from a mobile client is expensive due to the limited bandwidth of a wireless channel and the instability of the wireless network. We address this problem by maintaining a materialized view in a mobile clients local storage. Such a materialized view can be considered as a data warehouse. The materialized view contains results of common queries in which the mobile client is interested. In this paper, we address the view update problem for maintaining the consistency between a materialized view at a mobile client and the database server. The content of a materialized view could become incoherent with that at the database server when the content of the database server and/or when the location of the client is changed. Existing view update mechanisms are ‘push-based’. The server is responsible for notifying all clients whose views might be affected by the changes in database or the mobility of the client. This is not appropriate in a mobile environment due to the frequent wireless channel disconnection. Furthermore, it is not easy for a server to keep track of client movements to update individual client location-dependent views. We propose a ‘pull-based’ approach that allows a materialized view to be updated at a client in an incremental manner, requiring a client to request changes to its view from the server. We demonstrate the feasibility of our approach with experimental results.

Approximating object location for moving object database

We consider a moving object database, which keeps track of object locations and supports location-aware queries. Frequent location update to the database can provide high query accuracy but result in heavy consumption on the scarce communication bandwidth and precious object energy, as well as great index maintenance overhead at the server. We propose the use of safe-region: an area in which an object moves without having to send in a location update. The database records safe-regions for objects, instead of their exact locations for query answering and filtering. We propose three schemes to compute the safe-region based on object movement pattern and study the performance and the impact on query precision and cost.

Adaptive semantic data broadcast in a mobile environment

Permission to make digital or hard copies of part or all of this work or personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers, or to redistribute to lists, requires prior specific permission and/or a fee.