Stephen S. Rappaport

Traffic performance and mobility modeling of cellular communications with mixed platforms and highly variable mobilities

In modeling teletraffic performance of mobile cellular networks some characteristic mobilities are assumed to be known. It is useful if these assumptions impose few restrictions and lead to analytically tractable models. We present a new probability density function where the coefficient of variation can be larger than one but which nevertheless lends itself to analytical modeling using memoryless properties and multi-dimensional birth-death processes. This extends the previous framework to a broader class of mobilities. The approach allows computation of major teletraffic performance characteristics for cellular communications in which mobility issues are important. Multiple platform types and cut-off priority for handoffs are considered. Computational issues are discussed and some theoretical performance measures are obtained to demonstrate the method and compare with previous work.

Prioritized channel borrowing without locking: a channel sharing strategy for cellular communications

We previously suggested a new channel sharing method for cellular communications. The method, called channel borrowing without locking (CBWL), allows real-time borrowing of channels from adjacent cells without the need for channel locking in co-channel cells. CBWL with cut-off priority for calls that arise in the cell is presented. This scheme discourages excessive channel lending and borrowing at high traffic load and promotes a more uniform grade of service throughout the service area. An analysis using macro-states and decomposition is devised to evaluate the performance of the scheme. The results are validated by simulation.

A dynamic location tracking strategy for mobile communication systems

We present a dynamic predictive location management scheme using a continuous-time Markovian mobility model characterized by cell-to-cell transition probabilities. The size and shape of location areas (LAs) are determined dynamically and individually for each mobile user on the basis of gathered statistics and incoming call patterns. In this way, the combined average cost of location updating and paging signaling for each individual mobile user is reduced. In addition, we develop an algorithm to compute the probabilities of finding a mobile user in each cell of its current LA. An optimal multi-step paging algorithm is used to minimize the paging signaling cost. The numerical results demonstrate that the proposed scheme can significantly reduce combined location updating and paging signaling cost.

Mobile user registration in cellular systems with overlapping location areas

A family of registration schemes for location management in cellular systems with overlapping location areas (LAs) is considered. These schemes significantly reduce the location updating signaling traffic when compared with conventional (non-overlapping) schemes. This reduction is due to a decrease in the average number of mobile users who are in boundary cells of a given LA and who are registered to this given LA. We present an analytical model to evaluate performance. An algorithm to compute the average location update rate per user is developed. The analysis provides insight into the ability of overlapping LA structures to reduce the location updating signalling. Numerical results are obtained which show the dependence of average location update rate on the amount of overlap.

Dynamic base station selection for personal communication systems with distributed control schemes

Distributed control schemes allow base stations in personal communication systems to be placed at locations corresponding to high expected traffic. This flexible base station placement creates significant overlapping coverage areas that can be utilized to improve system performance [1]. A new technique for dynamic base station selection in systems with overlapping cells is considered and its effect on traffic performance is characterized. The technique realizes robust performance for personal communication systems in fluctuating and heavily tapered traffic. A mathematical analysis based on a state transition model is used to evaluate performance of a system that employs the proposed technique. The results indicate that improved blocking probability and carried traffic performance are obtainable. Computer simulations were undertaken confirm the analytical results.

High Capacity: Low Delay Packet Broadcast Multiaccess

Demand assigned packet radio schemes using satellites can achieve high bandwidth utilizations but they generally exhibit relatively poor delay performance at low traffic when compared with random access schemes. A modified demand assigned scheme is proposed which improves delay performance at low traffic by allowing random access transmissions by users who are waiting for channel assignments. The proposed scheme provides low average delays at low traffic (comparable to slotted ALOHA) and bandwidth utilizations comparable to that of demand assigned schemes. The scheme has potential application in situations where the traffic loading tends to vary widely.

Cellular Communication Systems with Voice and Background Data

Cellular communication systems that support both voice calls and background data calls arising from various platform types having different mobility characteristics are considered. Hand-off problems of voice and data calls are studied. A tractable analytical model for traffic performance is developed using multidimensional birth-death processes. Preemptive priority is used to assure transparency to voice calls. We examine two possible strategies for handling preempted data calls. One is a prioritized scheme and the other is non-prioritized scheme. Theoretical traffic performance characteristics are calculated. For voice calls, blocking probability and forced termination probability are calculated. The possible causes of data call termination are considered. These include forced termination probability of a data call and pushed-out failure probability. The expected waiting time of data calls is also determined.