Wireless multimedia streaming on hand-held, mobile or otherwise battery operated devices will be a major technology underlying the next generation information and entertainment appliances. From watching small video clips on your mobile device to playing live games with other people on the bus or looking at the latest news bulletin on your palm while waiting for your plane to arrive, the possible user applications are almost endless. The multimedia market traditionally consisted of closed (rather fixed) boxes, with a few battery-operated mobile devices such as the Walkman, but the market is now expanding its scope towards one where multimedia content can be retrieved from any place at any time. Our own home slowly becomes a networked environment, in which devices can talk to each other in either a wireless or wired manner, replacing the old stand-alone devices. But wireless multimedia streaming will also be possible when we are on the move, as so-called hot-spots, places in city centres, bus stations, airports, hospitals etc. come in place. Here our own devices can connect into a local networking area where services and content can be provided, which will give the possibility to retrieve the content we want and have a reasonable amount of bandwidth available.
It will be clear that timing requirements of multimedia systems change from a closed-box situation to the need for end-to-end timing specification and enforcement. Also when we look at a networked system consisting of multiple embedded devices, the question comes to mind how can we optimise the resource use, such as CPU power, network bandwidth, memory use and the power consumption over the complete system. As many wireless hand-held and mobile devices are short on resources and especially on energy, because they are battery operated, making trade-offs already during the design stage but also in an improvised way during their run-time becomes vital for optimal system performance and with that an optimal user experience for a reasonable cost. Today it is not possible, even at design time, to make well-founded system trade-offs between network and terminal resource consumption, energy consumption of the terminal and timeliness of the streaming data. This makes the quality of audio-video and gaming in the current first prototypes of wireless networked embedded devices by far not comparable to the high quality that people are used to from their traditional TV and audio sets.
The BETSY project will deliver all the theory, models and design methodology to make well-founded trade-offs between time-constraints, terminal and network resources and energy consumption possible during design time. The project will also deliver a framework implementation that makes dynamic adaptations in this trade-off possible at run-time. To verify the timing and resource model, the framework will be populated by selected components or modules, for example configurable codecs, bandwidth allocation controllers for WLAN cells and display drivers, which use the framework’s mechanisms to adapt the processing chain to changes in the resources mentioned above. The framework and its components will be implemented in a streaming server and mobile clients (or equivalent evaluation boards), and evaluated against the scenario definitions and their evaluation criteria. The possible saving of energy consumption in the terminal in a dynamic way compared with traditional methods will be measured as well. The expectation is that the terminal energy savings, depending on the situation, can be in the order of 20%.
The BETSY consortium has an excellent position to combine the research results of several domains, such as networking, device resource management, real-time processing and stream processing, as it consists of well-known academic and industrial partners in these fields. Philips, CSEM, IMEC, ISI, MDH, Siemens C-LAB, TU/e and University of Cyprus together aim to achieve a holistic view on the dependencies between bandwidth, delay, schedules, and the power and energy consumption for this specific application domain. We experienced in the past that much excellent research work was done in this area that nevertheless did not seem to fit in with the practical real-life requirements that CE and mobile devices had to meet. The BETSY project will avoid this by having two industrial partners Philips and Siemens, both European leaders in the CE and mobile domain combine efforts with best-in-class academics and research institutes. In this way, there is a balance between academic input, which is needed for the theoretical underpinning of the models and design paradigms as well as industrial input, which is needed for the practical exploitation of the results and the transfer of theoretical models into models usable in practice.
The results of BETSY will lead to reduced product cost by eliminating pessimistic and large safety margins or improved system performance with equal resource demands. The adaptive framework can reduce time to market of new applications as it facilitates integration of software components on handheld terminals, while guaranteeing a dependable system. The project results will also lead to automatic adaptation of devices and applications to the resource and energy limits of the surrounding environment, which is an important step towards Ambient Intelligence.