Wireless hop-by-hop credit-based flow control extended to source for stable best effort traffic

Abstract

Data traffic is expected to grow faster than capacity in future wireless networks. Therefore it will become unavoidable to deal with congestion. Bottlenecks are located on the wireless links because back-haul and Internet are overprovisioned. Traffic routed towards the user terminal (UT) in down-link direction keeps coming in through a big pipe until it reaches the base station (BS). The following wireless links can only carry a limited data rate due to congestion. In a multi-hop situation buffers before the bottlenecks ramp up and become unstable, leading to packet loss. While real-time traffic is safe due to call admission control (CAC), highest static priority and over-provisioning, best effort data traffic experiences congestion and therefore packet losses. A wireless flow control based on a credit-based hop-by-hop concept can solve this problem by avoiding any buffer overflow completely. This paper proposes extending the closed flow control loops to the source, either by a genuine credit-based flow control or by TCP rate control with deep packet inspection and ACK modification. This paper analyses the queueing behavior with stochastic Petri nets models. Markov state analysis provides numeric performance results. The example scenario consists of two wireless relayed hops and a wired back-haul with different control approaches for the hop between source and bottleneck.