Quantifying the Benefits of Exploiting User Profiles in Cellular Networks

The manner in which users cross cell boundaries in a cellular network has a high degree of predictability due to temporal and spatial locality in user mobility. This is so because user typically take predictable routes during specific times of the day, and these routes cut across cell coverage areas in an ordered sequence. This mobility information on a per-user basis, referred to as user profiles, can be maintained in Home Location Register (HLR) and Visitor Location Register (VLR) databases in the cellular network in order to provide improved Quality of Service (QoS) to cellular users. This QoS improvement can be achieved by exploiting per-user profile to reserve cell resources–particularly the channel bandwidth–prior to cell crossings thereby reducing the handoff blocking (dropout) probability. We consider a model in which the network provides a profile-based QoS at a premium to subscribed users while supporting non-profile users as well. We propose and study the performance of a simple channel-reservation algorithm in which profiled users are allowed to reserve channels prior to entering a cell. A key design issue is to study the impact of the reservation distance and the reservation granularity on the network resource utilization. Through simulation and analysis, we show that this simple algorithm yields significant improvement in the dropout probability of profiled users, and there is an optimal value of the reservation distance and the reservation granularity parameters which results in maximal improvement in the dropout probability of profiled and non-profiled users. Finally, we have developed a simple but approximate analytical model to study the benefits of profiling in cases when simulation might be time consuming. Keywords— User profiles, Channel-reservation algorithm, Quality of Service, Dropout probability, Simulation, Analytical model, Reservation distance, Reservation granularity.