The use of adaptive resource management is of particular interest for soft real-time applications, e.g., multimedia applications within consumer electronics systems and in telecommunications. Consumer electronics products range from miniature cameras and MP3 players to advanced media servers and large displays. Mainly driven by Moore’s law, the evolution in the CE industry is very fast. Utilizing available hardware and software resources in an optimal fashion is crucial both to save costs and to keep the competitive edge. Moreover, multimedia systems exhibit a highly dynamic behavior, since task execution times are often dependent on input data that are difficult to predict. As a consequence, these systems are prone to intermittent overload conditions that could degrade the performance in an unpredictable fashion.
The introduction of multicore platforms also in embedded applications creates new design challenges. A particular problem compared to uniprocessor platforms is the WCET analysis. Due to the shared memory access WCET analysis runs the risk of being very conservative. This will most likely hamper the application of hard real-time techniques based on static analysis. Hence, the market for more dynamic or adaptive resource management based on feedback from the true resource utilization and/or the application quality-of-service can be expected to increase in the future.
Another challenge, not only for embedded systems, but for all computing systems is the increased variability in circuit delays and power consumption caused by quantum effects in sub-45 nm chips. This causes classical device estimation models to fail and creates a need for variability-aware designs in which monitors and “variability knobs” can be used to dynamically adjust operating points.
Also in industrial sectors where predictability is the main concern there is always a certain need for adaptivity. For example, companies like Boeing have expressed a need for active resource management and dynamic scheduling as well as to handle, during system execution, things that were not anticipated at design-time. The automotive industry also has expressed increasing needs to handle software upgrades in a robust way and flexible attachment of devices (e.g. PDA’s, especially in the infotainment domain). In addition to resource management, this requires more stringent configuration management (be it on-line or support by off-line tools) that ensure that new configurations are compatible (both in a functional and non-functional sense).