Research On Spatio-Textual Publish/Subscribe In Road Networks

The development of Internet technology has changed the application and scale of distributed systems,which also brings higher requirements for the flexibility and scalability of such systems.The publish/subscribe system has been widely applied in many applications,such as banking,e-commerce and e-government,because of the merits of asynchronous transmission,multi-point communications and loose couplings.In a typical publish/subscribe system,event matching is the process by which a proxy server in a system checks for events posted by a publisher to find matching subscriptions based on each constraint.When there is a lot of subscription and publishing events,the proxy server is under pressure from the event matching operation,which may reduce the efficiency of the system.Therefore,the efficiency of the event matching algorithm is a key issue for large publish/ subscribe systems.There are still many problems with the existing publish/subscribe system.In the traditional publish/subscribe system,they only consider the matching of keywords,but neglect the spatial location information attached to the text.Second,many publish/subscribe systems that take into account spatial address locations are also limited to common European distances and can't handle the more realistic road network distances.Third,subscriptions and events in the system continue to occur and die over time,most of the systems are only suitable for dealing with static situations,ignoring the life cycle of subscriptions and events in the system.This paper focuses on the parametric spatial text publish/subscribe in the road network.Two index structures,Basic Index Architecture and Enhanced Index Architecture,are proposed respectively,to effectively deal with the parametric spatial text publish/subscribe problem.In addition,several reduction techniques are proposed to reduce the large number of subscriptions that are not relevant to incoming events in order to improve the processing efficiency.Based on the proposed index structures and pruning strategies,an efficient algorithm for event distribution is proposed.Finally,the proposed techniques are extensively evaluated with a real road network and two data sets.The experimental results show that the based on Enhanced Index Architecture method is more efficient and scalable than the based on Basic Index Architecture method.Finally,the paper gives the development prospect of the publish/subscribe system and puts forward further research directions on the matching algorithm.