Comparative evaluation of distributed physical cell identity assignment schemes for LTE-advanced systems

Abstract

One of the main challenges pertinent to Long Term Evolution (LTE) systems is the development of distributed and self-organized solutions for the Physical Cell Identities (PCIs) assignment problem. This challenge is the focus of the paper, which highlights the state of art and investigates via simulation studies the performance of three well-known distributed approaches (the graph coloring approach, the LTE standard proposal and the NS approach) deploying several scenarios, namely static, growing and dynamic. Our simulation results show that the NS approach outperforms the two other studied schemes in all investigated scenarios: it reduces collisions by more than 70% and confusions by more than 40%. We also show that the deployment of eNBs significantly affects the performance of the PCI assignment approaches. The network converges to a steady state employing all studied approaches deploying the static and growing scenarios. However, there is no guarantee to reach a steady state when deploying the dynamic scenario. In fact, this depends on the degree of dynamic the network has. The higher the dynamic degree, the higher is the probability that the system transits into an unstable state. Our results show that the probability of experiencing confusions when deploying the dynamic scenario is 0.45, 0.57 and 0.57 when operating the LTE proposal, the graph coloring and the NS approach, respectively. Our studies conclude that considerable developments to the approaches must be achieved to cope with dynamic scenarios.