Video Transmission for Third Generation Wireless Communication Systems

Journal of Research of the National Institute of Standards and Technology, March-April, 2001 by H. Gharavi, S. M. Alamouti

The uplink short and long scrambling codes are used to suppress intercell and intracell interference from the various mobile stations at the base station receiver. The mobile stations in a given cell are each assigned a unique short scrambling code. These short scrambling codes are from the periodic extended S(2) code family of length 256. Because of the large number of these codes, it is possible to have a unique code for each mobile station in a given cell and also to make sure that the mobile stations in the neighboring cells are assigned different codes. The long code used on the reverse DCDPCH/DCCPCH is 38 400 chips of a complex [2.sup.25] - 1 Gold Code. The quadrature component of the code is a decimated and shifted version of the in-phase component of the same Gold code. The quadrature component is also multiplied by repetitions of a 1 and -1 sequence to improve the envelope properties of the signal. The WCDMA has provisions for multi-user detection where a base station can detect multiple users at the same time, hence, increasing the effective signal-to-noise-ratio (SNR) of all the users. Only short scrambling codes are applied when multi-user detection is used at the base station receiver. Otherwise, only long scrambling codes are used.

3. Source Coding and Video Partitioning

Given a certain total bitrate budget, use of strong error-correction coding increases the error resilience of the video transceiver scheme at the cost of reducing the number of bits available for video coding. Similarly, it is possible to increase the power of the channel codec assigned to the protection of the perceptually most important video bits at the cost of reducing the protection of the less important video bits. However, it is intractable to directly optimize the associated partitioning. Furthermore, it is possible to assign the video bits to a high number of bit protection classes; however, in most practical cases, it is sufficient to employ two or three protection classes [13].

The partitioning process is, in general, described with the aid of the percentage of bits assigned to the individual partitions. The number of different bit-sensitivity classes and the video bits assigned to them has to be decided on the basis of the visual importance differences of the various video bits, although the rigorous formal evaluation of these sensitivities is an arduous and time-consuming task. Hence, in a somewhat simplistic, but plausible approach, it is often argued that in a subjective sense, the visual importance of the various video parameters is related to the spatial frequency of the video features described by the parameters. This is also true in the case of interframe coding, when the objects move at a high velocity, since the combined effects of high-velocity, high-frequency video contents cannot be accurately resolved by the human eye due to its relatively low so-called fusion frequency.

The above observation is inherently exploited in sub-band [12] and Discrete Cosine Transform (DCT) [15,161 based coding, since typically a lower proportion of bits is assigned to the high-frequency video components than to their low-frequency counterparts. Hence, the partitioning of their bitstreams is relatively straightforward. The so-called inter-frame hybrid DCT coding technique has been adopted for most practical video codecs, including the existing video coding standards [7,8,9,10,11]. The partitioning of the associated video reams has also received considerable attention in recent years; e.g., in the context of Asynchronous Transfer Mode (ATM) networks for mitigating the effects of cell loss [17,18,19], and for providing SNR scalability--a term indicating that different quality video streams corresponding to different bitrates can be generated--as provisions in the MPEG-2 [11] and H.263 standards [9].