A Novel Distributed Latency-Aware Data Processing in Fog Computing-Enabled IoT Networks
K. E. Srinivasa Desikan, Manikantan Srinivasan, C. Siva Ram Murthy · 2017
In generic Internet of Things (IoT) architecture, all the data generated is sent via gateways (GWs) and is processed in Cloud. This approach limits many real-time applications due to high propagation latency to the Cloud and also under-utilizes the GWs' compute and storage resources. Fog Computing (FC) extends the computability and storage of Cloud computing paradigm to network's edge devices such as GWs. IoT networks with FC enabled GWs, mitigate the latency and underutilization problems by processing data at the GWs. We propose a distributed latency-aware data processing (DLA-DP) model by which FC enabled GWs dynamically exchange processing and storage capability information and probabilistically forward data to its neighboring GWs or to Cloud only when there is a limit in local processing or storage. Modeled as a network of M/M/m/B queuing systems, a DLA-DP enabled IoT network is validated with extensive simulations. DLA-DP model enables improvements such as reduced system response time, increased gateway processing and buffer occupancy efficiencies.