Grain-oriented computer architectures for dynamically-reconfigurable avionics systems

Reconfigurable Computer Architectures (RCAs) still needs to be exploited since no many approaches fully squeeze their capabilities. Efficient allocation of computing resources in dynamic demand-driven deployments for avionics is all a challenge since they face costly and risky certification processes. This paper reviews the current RCAs with a strong emphasis on a full range of granularity and dynamic resource management. It has a good argument for having an effective control at the smallest reconfigurable RCA unit for efficient energy. The review of RCAs has a generic nature but aims to study RCAs for avionics self-adaptation. Unmanned Air Vehicles (UAVs) can be a first suitable test-bed to assess dynamically-reconfigurable avionics solutions. They are simpler than passenger aircraft, and currently required to endure much longer missions. Therefore, efficient power consumption is critical. The software-hardware computer interaction is essential to successfully implement optimization mechanisms to save energy. This paper describes an overview and a grain-based classification of RCAs. It presents a review that reveals gaps of current RCAs for avionics. A dynamically RCA is also proposed based on the flexibility of programming languages, and the reconfigurability of reprogrammable devices. Future research directions are presented as well.