I. Introduction The concomitant developments of the Internet, which offers to its users always larger and more evolved contents (from HTML (HyperText Markup Language) files to multimedia applications), and of wireless systems and handhelds integrating them, have progressively convinced a fair share of people of the interest to always be connected. Still, constraints of heterogeneity, reliability, quality and delay over the transmission channels are generally imposed to fulfill the requirements of these new needs and their corresponding economical goals. This implies different theoretical and practical challenges for the digital communications community of the present time. This paper presents a survey of the different techniques existing in the domain of HTML and video stream transmission over erroneous or lossy channels. In particular, the existing techniques on joint source and channel coding and decoding for multimedia or HTML applications are surveyed, as well as the related problems of streaming and downloading files over an IP mobile link. Finally, various diversity techniques that can be considered for such links, from antenna diversity to coding diversity, are presented. II. Basics of digital communications As depicted in Figure 1, a digital communication chain classically contains a source encoder, a channel encoder, a modulator, a physical channel and the converse operations [148]. The source encoder has for purpose to compress the digital signal that will be then transmitted over the physical channel of limited bandwidth, This critical operation consists of extracting most of the redundant information present in the signal, to produce a reduced set of symbols. These compressed symbols being extremely vulnerable to errors, which are likely to occur on the physical channel, the channel encoder is present to protect the compressed information stream and guarantee a sufficient level of quality to allow the reconstruction of the signal by the source decoder. The modulator role is to shape the protected signal for the actual transmission over the channel. Modern multimedia applications handle heterogeneous contents: images, audio and video sequences, web pages... These contents are characterized by a large redundancy in the data stream. In order to be able to efficiently transmit them over a limited physical channel and to respect real-time constraints, it may be necessary to reduce this redundancy in the coded stream. Different types of redundancy can be distinguished depending on the type of the considered content: spatial, temporal or statistical redundancy [170], and two different classes of source coders can then be employed. The first one corresponds to lossless techniques, which allow a compression without deterioration (see in the following, the case of HTML [2] compression by DEFLATE [43] over HTTP [53]), the second one to lossy techniques, which allow to reach higher compression rates at the cost of a slight degradation such as with H.263 [1] or H.264 [3]. More details on these source coding techniques and relative standards, as well as of channel coding principle and the corresponding Shannon theorems are presented in Section 2. However it is now well known that the hypotheses of Shannon separation theorem are rarely satisfied in practice, leading to research on joint techniques [76]. The constraints introduced by the presence of a network (IP) layer have also lead to study specific techniques of adaptation such as adaptive error correction for H. 264 [21]. III. Joint source channel coding and decoding robust techniques To offer a good Quality of Service (QoS) respecting the real-time streaming application constraints, different techniques of source robust coding and joint source channel coding/decoding can be employed. In particular one find multiple description coding (MDC), the description being obtained by sub-sampling [90][10], quantization [56][151][186], transforms [192][65][119] or even specific strategies such as forbidden symbol introduction [24][33][25][41]. Robust joint source channel coding techniques, aim at modifying the encoder in order to introduce redundancy in the coded binary stream [49], to include in particular the addition of markers [106], unequal error protection solutions [105] [78][126][21], or even a more general cross-layer approach [42][127]. Another interesting source channel coding technique, which optimizes the source encoder according to the channel is the domain of channel optimized vector quantization (COVQ) [54][55][154], Other solutions can be specific to given codecs, such as the work provided in [114] for the Lempel-Ziv codes. A third aspect of joint robust technique concerns the decoding side, to correct errors by exploiting residual redundancy left by the source coding operation, In particular, soft decoding techniques applied to variable length codes [128][134][200][165][108][114], or their arithmetic declination [80][66][69][71] is a field well documented, as well as the application of such decoding to specific standards [143] [20] [86]. IV. Diversity techniques for IP transmission in point/multipoint context In order to guarantee a given QoS, minimizing the length and frequency of the fading effect at the receiving antennas is critical. Three different diversity axes: temporal, frequential and spatial can be used to compensate the propagation of channel impairments, knowing that specific antenna diversity (polarization, angle...) can also be considered. Various strategies of combining different signals at the reception side [62] and theoretical bounds for the end-to-end distortion of MIMO channels have been established [81]. Diversity is also to be considered in the network itself, whose structure can be either centralized or ad-hoc. Whatever the used diversity strategy, the aim is to use it for taking advantage of robust source coding techniques like MDC or SVC (scalable video coding) [50][123][167]. Naturally, the diffusion strategy, which can be roughly classified into either Content Delivery Network (CDN) [112] type, in the case where servers are deployed, Peer to Peer (P2P) solutions for decentralized systems [95] or a combination of them [139], should be also considered and adapted. V. Conclusion The variety of techniques allowing to improve the quality of video or HTML transmissions over an IP wireless link is particularly wide. Different approaches, often additive, can be used to either better prepare the signal for the channel impairments, or improve the received signal by introduction of diversity (path, antennas...) or finally to improve the reconstruction of the degraded bitstream.
