Research On The Assurance Of The Service Quality Of Real-time Information Collecting Tasks

With the rapid development of the Internet, information of the Internet increasesrapidly. More people participate in the Internet to express their opinions andcommunicate with each other. In order to fully analyze and predict the trends of thepublic opinion, many related technologies have emerged. By the end of2012, there isa total number of more than60billion web pages in China, and remained100%growth rate since2003. Its explosive growth leads the growth of the public opiniondata by the exponential expansion of the number. Public opinion informationcollection must be done at a high speed in order to cope with the changes in publicopinion. Outdated information often miss the opportunities and lose value. Therefore,returning the collection of real-time information has important significance. The maincontent of this paper is that search requests from the processor scheduling andresource management problems by reptiles can be abstracted as a model in whichreal-time tasks and duplicated real-time tasks scheduling in heterogeneousmultiprocessors.Real-time tasks scheduling methods are mainly based on the traditionalsingle-processor scheduling methods such as rate monotonic scheduling, andmultiprocessor scheduling like local scheduling and global scheduling method andevolutionary algorithms scheduling. Each method has its advantages anddisadvantages and adaptation conditions. Moreover, during the real-time taskallocation process, this method did not consider the problem of energy consumption,even if the partition result is a feasible solution, but increasing energy consumption.By analyzing the limitations of the above methods, we study the real-time taskallocation methods and proposed an improved algorithm based on particle swarmoptimization in heterogeneous processors, increasing energy consumption constraints,satisfing the allocation of tasks at the same time reducing energy consumption.Meanwhile, as to duplicated real-time tasks allocation method, we present a linearprogramming algorithm, approximate dynamic programming algorithm andpolynomial time complexity algorithm to solve the problem, which has reached thepurpose of a more quickly result and increasing the posibility of finding a feasiblesolution.Through the experiments, we not only simulate on randomly generated real-timetask, but also on the real time tasks that are generated by real processors, theproposed algorithm can both reduce the running time and the energy consumption,improving the speed and quality of the solutions. In comparison with other existing algorithms we find that the improved PSO algorithm has certain advantages in thetime partition tasks and the number of feasible solutions, as well as on reducing theenergy consumption.Besides, the polynomial time complexity algorithm which solvesthe real-time duplicated tasks allocation has a good performance on reducing theexcution time and improving the ability of data processing.The method to the real-time task and duplicated real-time tasks allocation inheterogeneous multiprocessors environment problem can accurately allocate andprocess the tasks while reducing energy consumption. It has provided a method forthe requirement that returning the collection of real-time information in publicopinion and provide a reference on improving the speed of real-time informationcollecting and the accuracy.