Performance Enhancement of Video Delivery Services in LTE Networks

Abstract

LTE includes an enhanced multimedia broadcast/multicast service(eMBMS)

but delay-sensitive real-time video streaming requires the combination of efficient handling of wireless link bandwidth and reduced handover delays, which remains a challenge. The 3GPP standard introduces a Multimedia Broadcast and multicast service over a Single Frequency Network (MBSFN) area which is a group of base stations broadcasting the same multicast packets. It can reduce the handover delay within MBSFN areas, but raises the traffic load on LTE networks.

In this dissertation, we first presents an MBSFN architecture based on location management areas (LMAs) which can increase the sizes of MBSFN areas to reduce the average handover delay without too much bandwidth waste. An analytical model is developed to quantify service disruption time, bandwidth usage, and blocking probability for different sizes of MBSFN areas and LMAs while considering user mobility, user distribution, and eMBMS session popularity. Using this model, we also propose how to determine the best sizes of MBSFN areas and LMAs along with performance guarantees. Analytical and simulation results demonstrate that our LMA-based MBSFN scheme can achieve bandwidth-efficient multicast delivery while retaining an acceptable service disruption time.

We next propose to transmit the real-time video streaming packets of eMBMSs proactively and probabilistically, so that the average handover delay perceived by a user is stochastically guaranteed. To quantify the tradeoff between the perceived handover delay and the bandwidth overhead of proactive transmissions, we develop an analytical model considering user mobility, user distribution, and session popularity. Comprehensive simulation is carried out to verify the analysis.

On the other hand, hypertext transfer protocol (HTTP) based adaptive streaming (HAS) is expected to be a dominant technique for non-real-time video delivery in LTE networks. In this dissertation, we first analyze the root causes of the problems of the existing HAS techniques. Based on the insights gained from our analysis, we propose a network-side HAS solution to provide a fair, efficient, and stable video streaming service. The key characteristics of our solution are: (i) unification of video- and data-users into a single utility framework, (ii) direct rate control conveying the assigned rates to the video client through overwritten HTTP Response messages, and (iii) rate allocation for stability by a stateful approach. By the experiments conducted in a real LTE femtocell network, we compare the proposed solution with state-of-the-art HAS solutions. We reveal that our solution (i) enhances the average video bitrates, (ii) achieves the stability of video quality, and (iii) supports the control of the balance between video- and data-users.