Modeling adaptive rate video transmission in Wi-Fi MANET

Abstract

Broadcast scenarios, such as streaming HD video to one or many clients, seem as a natural fit for the wireless medium, and indeed the two couple well in satellite TV networks. The same kind of compatibility with broadcast may be expected from modern Wi-Fi networks featuring extended-range Access Points (APs). Yet, the 802.11 protocol was not designed with video broadcast in mind, and therefore lacks crucial broadcast mechanisms, most notably, dynamic rate-adaptation and efficient Forward Error Correction (FEC). For satisfactory Quality of Experience (QoE), it is important that the video stream reaches its destination in time and with a minimal amount of errors. Therefore, in live broadcast, retransmissions should generally be avoided, a goal that can only be accomplished using a reliable feedback mechanism that informs transmitters of current packet loss rates. In this paper we present a live video transmission model for Wi-Fi MANET networks that dynamically adapts the streams' FEC overhead, by utilizing the fact that in Wi-Fi MANET all nodes are periodically broadcasting beacon frames. By overloading these beacons with packet loss-rate information, the transmitting nodes are able to adapt the rate of FEC-encoded redundant packets, the transmission rate and the modulation scheme, without relying on Wi-Fi ACK frames that are costly and inadequate for broadcast scenarios. Moreover, using beacon frames that are part of the 802.11 standard eliminates the overhead associated with proprietary feedback packets in alternative models of broadcast over Wi-Fi. Thus, our suggested scheme enables Wi-Fi MANET networks to reliably transmit live video over multiple hops. Using NS-3 simulations we validate that our FEC adaptation model is applicative and efficient over one hop. Thereafter, we demonstrate via theoretical analysis the tolerable delay associated with the transmission of video over multiple hops using a pipeline model.