Task-Based Content Delivery at the Wi-Fi Edge

Task-based content delivery methods that combine Java fork-join with Java New file IO are proposed for Wi-Fi edge computing devices. Fork-join task-based processing framework performs better than traditional Java multithreading for parallel computation on multicore CPUs, especially for divide and conquer algorithms. New file IO for synchronous nonblocking socket channels and file channels, memory-mapped files and asynchronous socket channels combined with fork-join perform better than traditional Java synchronous blocking IO for sockets and file streams under highly unbalanced processing loads. Performance of proposed methods for throughput, heap memory occupancy and scalability are evaluated and compared with the traditional method in a Wi-Fi network. Fork-join combined with asynchronous socket channels and memory-mapped files performs best and significantly over the traditional method.