Reliable Service Execution Based On Probabilistic Moving Prediction In Mobile Wireless Sensor Network

With the increasingly development of mobile devices, the services they can offerare becoming more and more professional and single. When combine the servicesprovided by these devices organically, we can realize more complex functions. Theprocess above is called services combination and such service-oriented computingmodel is becoming a trend. This paperaddresses the problem of MANET(mobile adhoc network) where the complicated services are realized by the combination of thesingle services offered by the mobile nodes in the network. However, some problemswill appear due to the motion of devices, such as signals instability caused by thechanging of the distance between two devices and short-term services availableresulted from the devices’ autonomously moving out or in. Both the problems willbring the services combination in mobile environment great challenges. More accurateand efficient consideration and judgments in the process of combination are helpful tomake the final choice.The previous works on services combination are based on static wireless networkenvironment. The multiple comprehensive performance evaluation and effectivenessplay important roles in choosing services according to the sensed information of thecurrent environment. However, the existing studies have rarely referred to theavailability evaluation of the service provided in specific space and certain time andthe reliability of the whole services implementation process when the nodes are in astate of motion. The mobility of MANET is the source of the instability of manycombined services.Firstly, the reliability of the connection between the mobile nodes is evaluated.The removable nodes are classified according to their speed and directions. Secondlywe predict the position probability of the nodes. Lastly, the probable effectivepositions of the nodes will be calculated through integral approach. The Monte Carlosimulated approximate calculation method is helpful in the whole calculating process.To select process, the minimal time threshold condition to guarantee thereliability is further considered and the maximal time threshold as well. Twoalgorithms are given to select the services provided by users, the completely searchingapproach and the Pareto-dominated local k-th optimization method. When the laterone is selected, a satisfactory result can be obtained and the rapid expansion of theresult set can be avoided as well. Finally, we show some experiments to verify the reliability of the motion model,and the validity and necessity of the local top-k optimization algorithm as well.