MDP-IoT: MDP based interest forwarding for heterogeneous traffic in IoT-NDN environment

Internet of Things (loT) a vision, being built today, holds a new rule for future "anything that can be connected will be connected". IoT needs to support a multitude of heterogeneous objects extended with sensors, actuators, RFID's, etc. These "Smart Objects" need unique identification, autonomous data transfer and communication with other objects. Consequently, these unique requisites of IoT need a promising future Internet architecture as it mostly revolves around data. Furthermore, the existing host centric IP standards though advantageous, faces challenges like additional protocols for mobility, end-to end security while deploying it with massive loT applications. Named Data Networking (NDN) project is a new evolving data-centric internet architecture with innovative capabilities like caching, named data, security which mainly suits the specifications of loT thereby proposed to solve the shortcomings of IP. NDN traditionally supports a PULL based traffic and its stateful forwarding engine despite its skillful nature need some modification while designing for an loT system. In this paper, our foremost work is to classify and prioritize loT traffic and enable delay-intolerant applications with low latency, to retrieve Data efficiently. Next, we propose a Markov Decision Process (MDP) based Interest scheduling for loT traffic with varying priorities and measure the performance with different traffic probabilities. Our simulation results show that prioritizing and treating requests based on their traffic type can reduce network load by 30% thereby improving QoS in an IoT-NDN environment. The MDP-based loT model schedules' the Interest to the best interface efficiently reducing the RTI' values on an average of 20%-30% than conventional forwarding strategies. The incurred delay is 30% better than existing work and forwarding strategies. (C) 2017 Elsevier B.V. All rights reserved.