Zhuoyao Zhang

HFilter: Hybrid Finite Automaton Based Stream Filtering for Deep and Recursive XML Data

XML filtering applications are gaining increasing popularity recently. Automata are generally adopted to construct query indexes for evaluating large numbers of XPath queries over XML streams. Usually only shallow data are observed in existing approaches. How to process deep and recursive XML data with low memory limitation efficiently is still a challenging issue. In this paper, we propose HFilter, a Hybrid Finite Automaton (HFA) based stream filtering approach, to solve this problem. We introduce the basic two-tier HFA (lazy DFA tier and NFA tier) first, which realizes data prefix sharing and memory overflow control to improve the filtering throughput. Then an optimized three-tier HFA with an extra pre-expanded DFA tier is put forward, which significantly reduces the restarting cost of HFA after memory overflow. Experiments show that our approaches work more efficiently than existing ones.

Air Indexing for On-Demand XML Data Broadcast

XML data broadcast is an efficient way to disseminate semistructured information in wireless mobile environments. In this paper, we propose a novel two-tier index structure to facilitate the access of XML document in an on-demand broadcast system. It provides the clients with an overall image of all the XML documents available at the server side and hence enables the clients to locate complete result sets accordingly. A pruning strategy is developed to cut down the index size and a two-tier structure is proposed to further remove any redundant information. In addition, two index distribution strategies, namely naive distribution and partial distribution, have been designed to interleave the index information with the XML documents in the wireless channels. Theoretical analysis and simulation experiments are also put forward to show the benefits of our indexing methods.

Two-Tier Air Indexing for On-Demand XML Data Broadcast

XML data broadcast is an efficient way to disseminate semistructured information in wireless mobile environments. In this paper, we propose a novel two-tier index structure to facilitate the access of XML document in an on-demand broadcast system. It provides the clients with an overall image of all the XML documents available at the server side and hence enables the clients to locate complete result sets accordingly. A pruning strategy is developed to cut down the index size and a two-tier structure is proposed to further remove any redundant information. In addition, two index distribution strategies, namely naive distribution and partial distribution, have been designed to interleave the index information with the XML documents in the wireless channels. Theoretical analysis and simulation experiments are also put forward to show the benefits of our indexing methods.

A Novel Air Index Scheme for Twig Queries in On-Demand XML Data Broadcast

Data broadcast is an efficient way for information dissemination in wireless mobile environments, and on-demand XML data broadcast is one of the most important research issues in this area. Indexing XML data on wireless channel is critical for this issue since energy management is very important in wireless mobile environments. Previous works have focused on air index schemes for single path queries. In this paper, we propose a novel air index scheme that builds concise air indexes for twig queries in on-demand XML data broadcast. We adopt the Document Tree structure as the basic air index structure for twig queries and propose to prune redundant structures of the basic Document Tree indexes to reduce the energy consumption. Then we propose to combine all the pruned indexes into one which can eliminate structure redundancy among the indexes to further reduce the energy consumption. Our preliminary experiments show that our air index scheme is very effective and efficient, as it builds concise air indexes and supports twig queries without losing any precision.

Efficient Data Scheduling for Multi-item Queries in On-Demand Broadcast

Data broadcast is an efficient way of data dissemination in the wireless mobile environment. Recently, many researchers pay attention to broadcast scheduling for multi-item queries which is a more practical problem in this field. However, only a few works discuss the scheduling problem in on-demand environment. In this paper, we propose a heuristic approach called LRL for the broadcast scheduling problem. Performance analysis is put forward and simulation experiments show that our approaches achieve a comparative improvement compared with the existing methods.

Skewed Wireless Broadcast Scheduling for Multi-Item Queries

The wireless broadcast scheduling for multi-item queries is a research issue of great importance. There are several methods proposed in previous works. However most of these studies do not consider data replication in broadcast schedule which leads to poor performance when the access probability of queries is skewed. In this paper we propose a hybrid method which gives better performance for scheduling multi- item queries especially those with skewed query access probability.

On-Demand XML Data Broadcast in Wireless Computing Environments

In mobile wireless systems data on air can be accessed by a large number of mobile users. Many of these applications including wireless internets and traffic information systems are pull-based, that is, they respond to on-demand user requests. In this paper, we study the scheduling and indexing techniques of XML on-demand broadcasting. Our main contributions are 1) pre-processing the original XML documents by pruning the superfluous parts before broadcast scheduling; 2) indexing XML documents in the broadcast disk using DataGuides structure. Experiments show our approaches reduce both access time and tuning time significantly.

An Efficient Bulk Updating Method for Finite Automaton Based XML Filtering Systems

In structure-oriented XML filtering systems, approaches that use event-based parsing techniques and automatons are proved to have sufficiently high performance. Many of these filtering systems are based on Finite Automaton (FA). In this paper, we study the updating techniques for the FA-based filtering engine, which is the most important component of an XML filtering system and propose an efficient bulk updating method which considers common prefixes among the new arriving queries. Experiments show that our method provides significantly better scalability and performance when compared to existing methods.

Preemptive Scheduling for Multi-item Queries in On-Demand Data Broadcast

Scheduling for multi-item queries is a problem of practical importance in the field of data broadcast. Most previous works solve the problem under the pattern that at each broadcast moment, the server broadcasts a number of data items which constructs a broadcast cycle with fixed length. Within this pattern, if a query is not scheduled in the current cycle, it usually has to wait for quite a long time which prolongs the average access time. In this paper, we propose a preemptive scheduling method for multi-item queries in on-demand data broadcast. It breaks through the previous pattern. Simulation experiments show comparative improvements of average access time through our approach compared with the existing approaches.

A data partition based near optimal scheduling algorithm for wireless multi-channel data broadcast

Data broadcast is an efficient way to disseminate information to large numbers of users in wireless environments. The Square Root Rule (SRR) is the theoretical basis for the single channel broadcast scheduling. In this paper, we extend the SRR and propose the Multi-channel Square Root Rule (MSRR) for scheduling variable-length data with skewed access probabilities on variable-bandwidth channels. The theoretical optimal average access latency is also provided. However, this optimal value can not be achieved in reality. Based on MSRR, we provide a two-phase scheduling algorithm which achieves near optimal access latency. First data are partitioned and allocated to different channels according to MSRR. Second, different scheduling strategies are adopted on each channel according to the skewness of data subset allocated on that channel. Experiments show that the difference of average access latency between our results and the optimal value is below five percent in most situations.