Predictive resource allocation evaluation with real channel measurements

Mobile services, especially video streaming, has seen a rapid usage increase in recent years. Base stations (BSs) need to employ smarter and efficient resource allocation strategies to maintain high quality of service (QoS) to users at all the times. Predictive resource allocation (PRA), is one such novel scheme, in which BSs seek to anticipate the user demands and offer service to users in advance. As a result, the QoS can be improved, network load can be distributed over time, while at the same time offering efficient utilization of BS power. In location-aware PRA, the BS exploits location information of the users to predict the channel expected variations and adapt the BS resources accordingly. We evaluate PRA strategies based on an empirical study of the radio channel variation from measured location-aided channel radio maps using a smart phone. We observed that gains offered by the PRA scheme are highly dependent on user mobility patterns.