Research On Key Technologies In Design Of Embryonic Bio-inspired Self-healing Hardware

In recent years,with the rapid development of space exploration in our country,more and more spacecraft is being sent into space.The complex and harsh environment in space easily cause the failures of the electronic system especially digital circuit in the spacecraft.Maintenance is extremely difficult and even unreachable if a fault happened in the system.Consequently,improving the adaptability and reliability of digital circuit is the key to enhance the task reliability of spacecraft.Embryonic bio-inspired self-healing hardware(embryonic hardware,or Embryonics)changes the circuit structure,fault tolerant model and design method of traditional digital circuit.It provides a new solution for improving the task reliability of electronic system.After about twenty years of development,relative researchs on Embryonics have generated many achievements,but there is still a large gap between theory and engineering application in the fields such as routing capability,adware consumption,configuration data processing mechanism,self-organizing and self-healing algorithm,and fault detection.Therefore,in order to promote the application of Embryonics,this paper focuses on these key issues,and the main research contents and innovative works are shown in the following sections.Firstly,a design method of multi-layer Embryonics based on graph theory is presented.Aiming at complex routing process of typical embryonic hardware structure,inspired by features of hormone communication,we design a multi-layer embryonic hardware structure by combining neighborhood connections and network on chip,and then introduce its principle and self-healing capability.Aiming at large hardware overhead,a graph theory based design method of Embryonics is presented after showing the Embryonics' development process.The realizing method of self-organizing and self-healing based on functional flow graph and functional flow graph with redundancy is discussed.The multi-layer Embryonics improves the routing capability and self-healing capability.The graph theory based design method integrates the self-organizing process and the self-healing process,and it can reduce structure complexity of Embryonics,which means low hardware consumption.Secondly,a modular self-organizing configuration mechanism for Embryonics is presented.To satisfy the configuration requirements such as cellular differentiation or replication in Embryonics,referencing the Theseus mechanism used in ubichip,we present a self-organizing configuration mechanism for Embryonics,and introduce its principle and characteristics in detail.Its time and hardware consumption is discussed compared with the Theseus mechanism.Its application method is shown via configuring the multi-layer Embryonics.The presented mechanism provides a solution for configuration information's operation in Embryonics.It can also be used for configuring other reconfigurable hardware circuit.Thirdly,a novel data driven self-organizing fault-tolerant method and a fault detect method are presented.Aiming at the complexity of traditional self-organizing and self-healing process,we provide a data driven self-organizing fault tolerant method for applications whose functional flow graph doesn't have feedback by referring to self-organizing theory.Simulations based on Matlab are conducted to verify it self-organization and fault tolerant capabilities.The solder joints of pins connected with input and output cells easily break down while there is no method to solve this problem.So we presented a solder joints fault detecting method.The data driven self-organizing fault-tolerant method realizes automatic cellular differentiation,integrates the fault-tolerant capability into the self-organization process,and provides a simple and effective solution for self-organizing and self-healing.The solder joint fault detection method expands the fault detection scope of Embryonics.Finally,an experimental platform for embryonic bio-inspired self-healing hardward design is presented.Aiming to the lackness of experimental platform,starting from the needs of Embryonics' experiment,we build an Embryonics experimental platform.An Embryonics development system based on Xilinx K7 FPGA is designed.In order to meet the fault injection needs,we develop an FPGA fault injection system for providing failure to Embryonics' experiment in Embryonics development system.An experiment about PID controller is designed to verify the function of the platform,and the results show that the designed platform can provide supports for Embryonics' development and its functional verification.