Reliability-oriented clustering algorithm for service search in ubiquitous stub environments

Service search has been introduced to exploit heterogeneous resources of distributed devices on the purpose of supplying ubiquitous services in ubiquitous stub environments, especially in MANETs. However, due to the characteristics of infrastructure-less, devices' limited resources, and dynamic topology caused by the mobility of devices, service search faces great risk of failure. Usually, clusters are the main way of organizing the devices in MANETs. Therefore, an effective clustering algorithm is necessary to ensure the reliability of service search. A maximized reliability clustering algorithm (MRCA) is proposed. We present predicted battery supporting time, CPU computing power, connecting degree and predicted velocity of devices, select the best devices as cluster heads. We combine the four factors together using FAHP algorithm. The simulations show that the MRCA can prolong cluster headers and members' valid time, reduce the consumed energy in the cluster's life cycle. This proves that MRCA can improve the reliability of cluster.