Though current multimedia protocols can potentially support many different types of data (i.e., text, images, audio, video and graphics), the future of multimedia computing is tied to issues of network bandwidth and scalability. The outlook is that potential users of network multimedia will, for the foreseeable future, be diverse: from users with access to low-bandwidth connections (i.e., infra-red, cellular modems, ISDN), to users with access to very high-bandwidth connections (e.g., Gigabit networks). Advances in data compression will alleviate bandwidth problems -- by effectively increasing bandwidth -- in low-bandwidth networks; but based on technology's experience with the insatiable demand for raw computing speed (since applications seem to grow without bound) and raw bandwidth (since network traffic continues to grow at an astounding rate), we expect compression techniques to be important for high-bandwidth networks as well. The problem of scalability requires solutions at the level at which protocols and processes interoperate.
As long as computing speeds continue to outpace network speeds, data-compression is going to be an invaluable commodity. When the network is slow relative to the computer, the computer can spend cycles on compression to reduce its demand on the network. At present this scenario is apparent with ISDN (64 Kbps per channel), twisted pair, modems (e.g., 28.8 Kbps for telephone, 4.8 Kbps for cellular), and in general, with shared networks. However, apart from strategies for compression alone, another important issue is to integrate data compression with protocol support for distributed computing. For example, traditional network protocol implementations are single-threaded and are thus unable to fully exploit interleaving of communication and computation. On the other hand, for data compression to effectively use the network, it is crucial that a compression system be able to overlap compression cycles with communication activities by a protocol system. Moreover, since our use of multimedia compression includes the context of isochronous applications (such as video and audio) with real-time constraints, we must employ a protocol runtime system that can deliver compressed data with quality of service (QoS) guarantees.