Download now Free registration required
With the development of CPU architecture and virtual networks, it is urgent to propose algorithms which can support lots of virtual routers on a multiprocessor system, and solve the problems of load balance of processors and stable high-throughput of routers. This paper builds an evolutionary game model and proposes a Reinforcement Learning Algorithm for dynamic multiprocessors selection. Reinforcement learning algorithm is designed with the ability to learn the utility information of other virtual routers in the same or different populations for processor-selection.
- Format: PDF
- Size: 1218.56 KB