Reconfigurable Computing for Smart Vehicles

The future of the automobile industry will be defined by the capacity to process large volumes of data originating from a multitude of interfaces in a timely manner. Presently, high-end vehicles have more than 100 electronic control units (ECU) performing multiple functions in parallel such as powertrain, chassis & safety, comfort, and infotainment. The components of this complex system communicate with each other via multiple network interfaces. As the trends toward advanced driver assisted systems (ADAS) and from in-vehicle communication to vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, exponential growth will be seen in the number of electronic control units used in an automobile. This growth will lead to a rise in fast, time-critical, and real-time data processing demand from multiple interfaces such as camera interfaces, sensors, other vehicles, and roadside infrastructure. In the automotive domain, data processing is time-critical and must be reliable. This system will demand high level of parallel, time-critical and fault-tolerant computing. Current multichip or software-based solutions may not be able to satisfy this demand. FPGAs are suitable for such applications as they are programmable and customizable, and can process high volumes of data in parallel on a single chip. Simultaneously, an FPGA can have sophisticated error correction algorithms providing reliability at the hardware level to provide the throughput required for automotive applications.