Toward network function virtualization for cognitive wireless mesh networks: a TCP case study

Abstract

Nowadays, cognitive radio (CR) is considered the key technology to deploy wireless nodes that are able to adapt their transmission parameters based on the characteristics of the environment. In our previous works, we leveraged the reconfigurability properties offered by the CR technology to implement Cognitive Wireless Mesh Networks (CMNs) operating on the 2.4-GHz industrial, scientific, and medical (ISM) band. Urban-X, a new architecture for Multi-radio Cognitive Mesh Networks, self-configures and self-adapts to the dynamic interference conditions experienced on the Wi-Fi channels in dense urban environments. Urban-X estimates traffic load of existing users outside the mesh in each channel and utilizes less occupied channels opportunistically. In such dynamic environment, a transport layer protocol can suffer from variation in delay and available bandwidth due to spectrum sensing and mobility. Several TCP solutions for the CWMNs have been proposed but they could not be employed for implementation due to complexity and inefficiency. Instead, we adopt network function virtualization (NFV) technique which allows to instantiate virtualized network functions (VNFs) dynamically on demands. For this, we propose overall architecture of NFV framework for the CMNs and apply TCP accelerators (ACC) VNFs to improve TCP throughput as a case study. Simulation results show TCP-ACC VNF can achieve a notable performance enhancement compared to previous TCPs for the CMNs.