Wireless Video Demonstrator

A wireless video demonstrator is utilised for developing

• video encoding and decoding, together with
• wireless QoS control protocols and functions

for real-time services over different network technologies. The demonstrator combines the functions of lower layer network protocols and higher layer adaptation for enabling an uninterrupted service. QoS functionality is emerging into wireless networks, based on the advancing development work on telecommunication standardisations. The support for QoS over separate links is first required, while the adaptation of the different links together for establishing and end-to-end service is the next critical element.

Reference Model

The functional reference model of the video demonstrator is presented in Figure 1. The architecture consists of three main layers: video encoding/decoding, video adaptation with Video Control Protocol (VCP), and link/network layers. The reference model has been introduced, as the actual implementation of the demonstrator alternates according to the operational environment and used network technologies. Consequently, there are a number of different networking technologies combined for video delivery, each technology containing its own characteristics and configurable parameters, such as network type specific QoS. The purpose is to manage heterogeneous networks in order to provide an end-to-end service.

The middle layer has been divided into two sub-layers for differentiating the video contents related functions (video stream adaptation) and video transfer functions (VCP). VCP centralises the control on a heterogeneous delivery system. The video encoder/decoder layer is independent of other layers, thus enabling the use of existing standard encoders and decoders. Similarly, standard link or network technologies may be directly utilised for video transfer.

Implementation Architecture

The architecture of the current demonstrator is presented in Figure 2. The demonstrator is a client-server system, in which the connection between the entities can be realised using different technologies. The client is referred to as video receiver and the server as video sender. Video receiver is a terminal with varying processing capabilities and network connection types.

The video encoder produces an encoded video stream from a raw digitised video input. Real-time video capture for encoding can be used, and has been implemented in the current demonstrator. For video encoder/decoder layer, the basic functions are to adapt a stream at the sender and conceal occurred errors at the receiver end. The quality of the decoded video is measured by frame loss and by computing the Peak-Signal-to-Noise-Ratio (PSNR) against the original video material. Different video adaptation tools will be developed and tested with the demonstrator. The encoded video stream is prepared for the wireless transfer by the video stream adaptation and VCP layers. This layer processes the video stream for robustness against wireless link errors. In addition, a link simulation module has been implemented for simulating video transfer locally. Different demonstrator system modules provide User Interfaces (UI) for the configuration and collecting of measurement data.

The central connective control module is the EncoderIO at the sender and DecoderIO at the receiver. These modules connect the different components together. The sender and receiver operate in Windows NT/2000/XP computers. Network Interface Cards (NIC) are attached to a PC through device drivers that operate under the legacy TCP/IP protocol stack. Over network drivers, the Winsock2 Application Programming Interface (API) is applied for applications. The Video Sender and Video Receiver modules that implement most of the video adaptation can also connect to NIC device drivers through tailored drivers, without any specific protocol stack. This type of connection is used for example in video transfer over Bluetooth.

Video Encoder Implementation

The video input for the demonstrator is acquired by either real-time video capture or by using test sequences stored on the local disk. The stored video can be either pre-encoded H.263 stream or raw video. The source video is digitised using Matrox Meteor II video capture card for CIF or QCIF formats. The video encoder is a parallel implementation of ITU-T H.263 v1 on a Hunt Engineering Hepc8 platform that includes four TMS320C6201 DSPs. With the high capacity multiprocessor platform, frame rates exceeding 30 fps can be provided also for CIF. The real-time video decoder is an error robust and proprietary implementation of a H.263 decoder. The platform is shown in Figure 3.