Transient flow level models for interference-coupled cellular networks

Modern concepts for cellular networks, e.g., heterogeneous networks and small cells, increase base station densities to satisfy the capacity demand and, hence, lead to a highly dynamic, interference-limited regime. For instance, base stations can be turned on and off dynamically in order to adapt the energy consumption, and frequency usage, to fluctuating traffic demand. The time scale of such operations depends on hardware and system capabilities, but it can be in the range of seconds or minutes. Moreover, due to frequency reuse, user data rates are mutually coupled via inter-cell interference. In order to manage and optimize such dynamic networks, intelligent algorithms and sophisticated system models are needed. In this paper, we focus on flow level models based on queueing theory. Existing models often assume stationary conditions (i.e., fixed traffic rates and steady-state), which may be inadequate for dynamic systems with time-varying arrival intensities. Addressing this issue, we derive transient flow level models that consider time-dependent, dynamic network behavior. Since the flow level model of interference-coupled queues renders analytically intractable, we propose different approximation techniques, e.g., aggregation of variables and average interference, and determine first and second order bounds as well. Numerical studies compare the accuracies of the different approaches, and confirm that transient effects are not negligible in interference-coupled cellular networks.