Agent Based Solution For Intelligent Location Based Service

With the rapid spread of globalization, the metropolises enlarge their scale duratively. As a result, persons will always be in a completely strange place. So, obtaining location-based information is of great importance to each of us. However, asking others for help when needed is rather boring and inconvenient. Retrieving information in advance can't meet the constantly changing demands either. Therefore, we more and more require some personalized, customized services which are location based to facilitate users, namely location based service (LBS).At present, the dramatic development in wireless communication technology and prevalence of mobile devices furnish us an excellent opportunity to solve the problems mentioned above. Nevertheless, the existing LBS solution has many deficiencies. For example, it can't communicate well in isomerous networks. What's more, it isn't able to notify mobile users accurately. All these issues turn into a bottleneck, which prevents LBS from further popularity. So, in this thesis, a novel LBS solution—E-LBS is proposed that can solve the majority of existing problems.In the thesis, we first give a brief introduction to the overall architecture of LBS system, indicating its disadvantages. Then we analyze the reasons why mobile Agent is more suitable for wireless communication. Second we introduce mobile Agent to E-LBS, which is to be the basic communication carrier. Subsequently questions about security and privacy are discussed in detail and the prerequisites that Agent needs are particularized and a complete comparison between Agent and RPC is betaken. Furthermore, a user behavior prediction model is proposed to deal with the low notification accuracy problem in traditional subscribe/publish system of LBS solution. In this model, we take the contextual information of mobile users into consideration and create several probability distribution tables. Then Bayesian network inference is introduced to make the notification more accurate. Last but not least, we make a presentation of partially implemented system and future work.