The mobile nodes in adhoc networks have the various movements. These movements break the links between nodes in network. This is a major problem because nodes don’t have any aware of link expiration time. If nodes of network know awareness of link expiration time they can choice a reliable and durable paths and save in time and energy because they don’t need to discover new paths. So we can predict the link expiration time and save the result in a prediction table. For mobility prediction in urban area we should obey a mobility model. For simulation of mobility prediction in an urban area we select the Manhattan mobility model. After computing the prediction table base on Manhattan mobility model we introduce a new routing algorithm that will improve the throughput of network compared with shortest routing algorithms.