Research On Location-Aware Publish/Subscribe Methods

The popularity of intelligent mobile devices equipped with GPS receivers has caused widespread interest in geo-textual publish/subscribe systems.Thus geo-textual publish/subscribe is used in many mobile network scenarios,such as mobile social media,geo-based product recommendations,etc.In a geo-textual publish/subscribe system,subscribers register their interests as geo-textual subscriptions.Then the system quickly and accurately sends information to the relevant subscribers.The information is published by publishers(i.e.,data owners,data providers,sellers,etc..Unlike traditional publish/subscribe systems,geo-textual data includes a geographic description and a textual description that are continuously generated and propagated in the form of a dynamic data stream.Geo-textual data has strong position sensitivity,for example,when a subscriber leaves a geographic area,the geo-related publication data is no longer sent to the subscriber.Therefore,higher efficiency and flexibility requirements have been proposed for more and more application scenarios.Around the issue of location-sensitive publish/subscribe,three main aspects of research work have been carried out: adaptive Top-K location text publish/subscribe method,semantic-based location text publish/subscribe method,and location-based distributed publish/subscribe architecture.For the Top-K location text publish/subscribe problem,most existing publish/subscribe systems not only need to match all publications for each subscription(which reduces the matching efficiency),but also need subscribers to pre-set appropriate publish/subscribe matching threshold(which reduces the flexibility of the system).To solve the above two problems,an efficient and flexible Top-K location-sensitive text publishing and subscribing method EPTG is proposed.This method classifies publishing and subscribing,so that when publishing and subscribing matches,only the same publishing is matched for the subscription,which greatly improves the matching efficiency.At the same time,an adaptive location-aware text publish/subscribe matching algorithm is presented.It does not require the user to pre-set the matching threshold,and can acquire Top-K location-aware publications for each subscription adaptively,which improves the flexibility of the system.Existing location-aware publish/subscribe systems mainly focus on spatial text data.Most of them directly regard keywords as simple text characters,often ignoring the specific semantics expressed by subscribers,so they can not return matching results more in line with subscribers' intentions.In this paper,a concept of semantic-based publish/subscribe for spatial and temporal text is proposed,considering three factors: At the semantic level,a document model based on potential Dirichlet assignment is proposed for semantic matching.Compared with the method based on text similarity,it can provide a more user-friendly query structure.In terms of spatial and informativeness,two indexing schemes based on maximum possible distance and conditional influence region are proposed.Efficient index matching for spatiotemporal text subscription based on semantics.Most of the existing location-based publish/subscribe solutions solve the scalability problem vertically.That is to say,the centralized computing unit responsible for the complete workload is used to calculate matching.Obviously,the vertical scalability method can only achieve a certain load,while the exponential growth of the ecosystem like the Internet of Things can easily exceed its load capacity.Therefore,the location-based distribution is proposed.Publish/Subscribe System Architecture,we use the concept of map tiling to segment the whole spatial locality,and combine the communication mode of message routing to dynamically allocate the computing process to complete the matching.We propose a priori conflict resolution protocol to ensure the computing unit's computing conflict when tiles overlap in distributed environment.At the same time,we adopt a strategy based on active replication to guarantee the system.Fault tolerance and reliability,all computing units in the architecture can be scaled horizontally.