Infrared and Laser Engineering, Volume. 45, Issue 11, 1126003(2016)
Resource-constrained parallel information processing methods based on reconfigurable technology
A method of reconfigurable parallel information processing in GNC was proposed for conflicting issues between high calculation performance and low power consumption in information processing of cubesats. An architecture of tightly coupled reconfigurable parallel processing was adopted in the method and complex algorithms which require many iterations and are not suitable for CPU operations in GNC information processing were achieved by dynamically partial reconfigurable unit of the hardware circuit (DPR). A multi-core parallel reconfigurable resource scheduling algorithm based on mutex was adopted in the method and optimized hardware acceleration of the software algorithms was completed through multi-core CPU parallel calls by DPR unit. The experimental results indicate that real-time information processing of GNC is effectively achieved in cubesats and reduces power consumption up to 50%. The method can be widely used in the field of imformation processing for resource-constrained satellites and possesses a great application prospect.
Get Citation
Copy Citation Text
Lu Zhijun, Ben De, Mao Bonian. Resource-constrained parallel information processing methods based on reconfigurable technology[J]. Infrared and Laser Engineering, 2016, 45(11): 1126003
Category: 信息获取与识别
Received: Mar. 15, 2016
Accepted: Apr. 18, 2016
Published Online: Jan. 20, 2017
The Author Email: Zhijun Lu (luzhijun2004@sina.com)