Maximizing the capacity of mobile cellular networks with heterogeneous traffic

We propose a novel adaptive reservation scheme that handles, in an integrated way, heterogeneous traffic of two types: streaming and elastic. The scheme adjusts the rates of streaming sessions to meet the QoS objective, adapting to any mix of traffic and enforcing a differentiated treatment among services, in both fixed and variable capacity systems. The resource utilization achieved by streaming traffic is close to the one obtained by an optimal policy, while the efficiency in the use of resources achieved by elastic traffic is greatly improved by limiting the abandonment probability. The performance evaluation carried out verifies that the QoS objective is met with an excellent precision and that the scheme converges rapidly to new operating conditions. We also compare the new adaptive scheme with two previous ones verifying that ours performs better in terms of carried traffic and convergence rate. The proposed scheme has low complexity which makes it practical for real cellular networks.