Thomas Stockhammer

System Design and Advanced Receiver Techniques for MBMS Broadcast Services

Raptor codes have been recently standardized by 3GPP to be used in the application layer. These codes are very efficient and perform sufficiently close to the so-called ideal fountain code on the block-erasure channel. However, for real applications, a simple receiver enhancement called permeable layer receiver can be done, which provides huge performance gains. Standard Raptor decoding algorithms have to be revised and new efficient decoding algorithms need to be introduced. In this work we present a practical and efficient implementation of the Raptor decoder for permeable layer receiver. We show that, not only we achieve huge performance gains, but we achieve them at an affordable (low) decoding complexity. Complexity and decoding performance simulation results are presented for the download delivery and video broadcasting.

Cross-layer assisted reliability design for wireless multimedia broadcast

In this paper, different error protection strategies in different protocol layers are discussed, in order to reliably broadcast multimedia data within existing wireless networks. In particular, we investigate the performance of additional forward error correction (FEC) in the physical layer, the radio link control (RLC) layer, and the RTP layer for multimedia broadcast and multicast services over GERAN systems. Advantages and drawbacks when applying error correction in different layers are shown. We introduce a simple receiver modification, referred to as permeable-layer receiver (PLR), which exploits traditionally useless information at the receiver, while the transmitter is kept unchanged. Significant performance gains are reported. Finally, the application to H.264/AVC based wireless video broadcast is discussed and the performance of different system designs for video transmission is shown.

Enrichment of speech calls by live video

This contribution addresses the case when live packet-switched video is used to enrich circuit-switched speech calls in mobile telephony. Circuit-switched and packet-switched transmissions generally operate on completely different transmission paths resulting in different QoS in terms of delay and loss rates. In this case, video and audio data recorded at the same time are not multiplexed together and might arrive at completely different time instances at the receiver. This is a major challenge for the receiver and the service as lip synchronicity between the voice and the video can generally not be expected. Our presented method solves the synchronization problem purely by the use of signal processing, i.e. no inclusion of time stamps is needed. The method is simple to implement and it is able to handle signal disturbances over the link.

Iterative decoding for GERAN MBMS

Various advanced methods have been proposed to achieve sufficiently high throughput as well as reliability for multimedia broadcast multicast services (MBMS) within 3GPP RAN and GERAN. One of the methods which yield promising results with only marginal system modifications is based on outer Reed-Solomon (RS) coding on the radio link control (RLC) layer. For initial studies standard decoding algorithms based on erasure-only decoders have been employed. In order to further improve the performance, the system can be viewed as a serially concatenated code such that iterative decoding can be performed. Then, advanced receivers for MBMS over GERAN would incorporate soft-in soft-cut (SISO) decoders for both the inner convolutional code, as well as the outer RS code. Whereas for the convolutional code, standard SISO algorithms are applied, for the outer RS code a standard Berlekamp-Massey algorithm is modified to accept soft-input and to provide soft-output. With this modified iterative decoding strategy, simulations have shown gains of up to 14% in throughput on typical mobile radio channels