Real-time combustion knock processing using a single instruction multiple data automotive power PC system-on-a-chip

This paper discusses a novel high performance knock processing strategy using a next generation Motorola automotive PowerPC system-on-a-chip. The proposed methodology is based on an auxiliary signal processing extension to the main PowerPC system-on-a-chip core along with various intelligent autonomous on-chip modules. Real-time software development techniques with an advanced software circular buffer implementation for processing the streaming knock sensor data have been developed. Various single instruction multiple data software optimisation techniques are employed to reduce the real-time knock algorithmic execution time. Real-time and simulation results are presented for the detection of knock on a four cylinder internal combustion engine, however, the approach is widely applicable. The efficient coding and optimisation techniques used for the single instruction multiple data implementation of the algorithm have been shown to improve overall performance and as a result utilises minimum combustion event timing.