Cross-layer multiuser session control for optimized communications on SDN-based cloud platforms

摘要

The plethora of emerging applications and their increasing popularity has led to the cloudification of the Internet of Things (IoT). The IoT-cloudification scales up data centers to a high computational resource capacity, to suit the requirements of a large number of users wanting to access multiple services offered by revolutionary IoT applications which, in turn, consume huge amounts of data produced by billions of things. In such systems, IoT-tailored cloud services (e.g., software platforms and middleware) are responsible for running latency-critical IoT applications in different areas (e.g., healthcare, transport systems, safety, Smart Cities, and many others). These applications expect to produce high-density multimedia flows that must be handled appropriately by the underlying communication infrastructure to obtain transport services of the highest standard. In this work, we propose a novel control plane optimization, which leverages the support of the Software-Defined Networking (SDN) substrate so that it can advance beyond today’s IP-based communication systems, which reveals weaknesses (e.g. a limited performance and scalability) in delivering multimedia content through per-flow channels. An attempt is made to overcome these limitations by proposing CLASSICO, a Cross-LAyer Sdn SessIon COntrol approach that exploits the SDN substrate to offload the flow streaming network-computing task from the IoT cloud platform to selected SDN branching nodes, and hence ensures high timeliness and scalability for the IoT-cloud system as a whole. To achieve this, CLASSICO dynamically builds application-layer sessions, which comprise multiple participant members sharing same interested content, and connects them into optimal multiuser tree structures, whereby only selected branching nodes apply SDN replication function for quality-enhanced transport service. We applied CLASSICO to a multimedia use case, and the results show that it outperforms multicast-capable IoT systems in terms of Quality of Service (QoS) and Quality of Experience (QoE) video assessment metrics.