System Modeling And Verification Of An Eye Gaze Tracking SoC

Studies have shown that human information processing is largely dependent on the vision,about80%-90%of the information from the outside world are obtained through the eyes ofthe people.Gaze tracking system can obtain human visual perception through tracking thehuman eye gaze point,and it has broad application prospects in various fields,such ascommercial, military, entertainment, life and so on. With the rapid development of theintegrated circuit size and manufacturing technology, SoC(System on a chip), which aims atmultidisciplinary cross connection, is becoming the main solution of many electronic products.And the following research of the SoC design methodology also is gradually being theconcern, however, the immaturity of the SoC design methodology provides us with a lot ofresearch and development space.By comparing the traditional SoC design methodology, the paper, take the SoC which isbased on Purking image appoach as an application, use Electronic system level designmethodology to design and verify in SystemC model, along with conver the SystemC modelinto the HDL model manually to finish the verification on FPGA, which is the actualhardware platform. The content of this paper mainly include the following parts:Firstly, this paper identify all the algorithms in the eye gaze tracking system throughdeeply analyzing their current research status, along with the experiment. In addition, thispaper compare the number of performance factors in different layer model of System designlanguage-SystemC, and make sure that Transaction level model is the higher abstract layermodel which is used for partitioning software and hardware of this SoC. And after that, thehardware will be refined to register transfer level model.Secondly, the eye gaze tracking SoC is modeled at high abstract level with TLM, and thesoftware and hardware partition result can be obtained from the simulation of TLM. So thepart of hardware is refined to RTL model, moreover, the hardware architecture can beoptimized with the simulation result of RTL mode. On the other hand, the part of software isaccurate timing simulated with the Instruction Set Simulator of embedded micro-processorNios II. So that the whole eye gaze tracking SoC is modeled and verified on the Computer.Finally, the hardware architecture after optimized and the software code after verified, are transplanted to hardware platform FPGA, using the manual SystemC-to-HDL method, andbuild a eye gaze tracking SoPC(System on a programmable chip) to complete the hardwareand software co-verification.The experiments show that the simulation result of the SystemC model and the simulationresult of hardware platform FPGA, are98%similar, that is saying that the SoC designmethodology based on SystemC model is efficient. And the SystemC modeling without thehardware, high abstract fast simulation, automatical software and hardware partition and so on,they are indicating that this paper’s SoC design methodology is fast and reliable.