Skip to Main Content
A location management scheme in wireless networks must effectively handle both user location update and search operations. Replication and forwarding are two well-known techniques to reduce user search and update costs, respectively, with replication being most effective when the call to mobility ratio (CMR) of the user is high, while forwarding is most effective when the CMR value is low. Thus, based on the user's CMR, the system can adopt a CMR threshold-based scheme such that if the user's CMR is lower than a threshold, then the system applies the forwarding scheme; otherwise, it applies the replication scheme. Applying different location management schemes based on per-user CMR values introduces undesirable high complexity in managing and maintaining location- related information stored in the system as different system support mechanisms must be applied to different users. In this paper, we quantitatively analyze a hybrid replication with forwarding scheme that can be uniformly applied to all users. The most striking feature of the hybrid scheme is that it can determine and apply the optimal number of replicas and forwarding chain length on a per-user basis to minimize the communication cost due to location management operations while still being able to use the same data structure and algorithm to execute location management operations in a uniform way for all users. We develop a stochastic Petri net model to help gather this information and show how the information obtained statically can be used efficiently by the system at runtime to determine the optimal number of replicas and forwarding chain length when given a use user's profile. We show that the proposed hybrid scheme outperforms both pure replication and forwarding schemes, as well as the CMR threshold-based scheme under all CMR values.
Date of Publication: Jan.-March 2003