Research Of General-purpose Signal Processing Technology Based On Cloud Theory

Cloud computing is an emerging computational model and is also one of the hotresearch topics in domestic and international research institutions and major companies inthe world. Virtualization and automated management of computing resources is animportant feature of cloud computing. In cloud computing environments, resources likeprocessors, storage and networks are interpreted as a resource pool by single-systemvirtualization technology, which can form powerful computing power through unifiedscheduling and management by software of automatic management. With advances in radartechnology, radar signal processing system has gradually made a transformation fromtraditional simplification and specialization to trend of universalization, standardization andmodularization. High real-time of processing of large amounts of data requires using ofdistributed parallel system, which brings a challenge of task allocation and management inradar public information processing platform with the number of sub-systems increasing, soa new supervisor mode to meet the requirements of automation is in urgent need.This paper discusses the background and development status of cloud computing inmore detail, analyses technology architecture of cloud computing by comparing domesticand international existing cloud computing platforms, researches the feasibility of applyingcloud computing to radar signal processing by analyzing of technology components andhigh real-time, high-bandwidth characteristics of radar signal processing platforms, thendesigns a signal processing cloud model composed of peripheral tree subnet and corenetwork and its software architecture, discusses the subnet composition and its mainfunction which is based on the open source cloud computing platform Eucalyptus, besides,builds a private cloud platform after simulation of core network’s topology designalgorithms. In aspect of resource management technology, this paper analyzes the needs,strategy, advantages and disadvantages of common load balancing scheduling algorithms,and simulates a cloud computing center by using simulation tool CloudSim, at the sametime, gets a conclusion of that greedy algorithm is able to get the local optimal solutionwhich possesses the property of correctness, validity and a simpler implementation in cloudcomputing resource scheduling by comparing with Round-Robin algorithm of Eucalyptusdefaulted, which provides a basis for the expansion of Eucalyptus. Finally preliminarilyresearches technical advantages of signal processing cloud by combining with working mechanisms and guarantee of real-time services.