Game Theoretic Congestion Control to Achieve Hard Reliability in Mission-Critical IoT

Mission-critical Internet of Things (MC-IoT) applications span from the industrial field to the battlefield and from smart homes to healthcare. Reliability is one of the stringent requirements of such applications. Routing in low-power and Lossy Networks (RPL) is a standard routing protocol traditionally used in IoT applications. Unlike the traditional single-path-based RPL, the reliable multi-path RPL (RMP-RPL) selects k parents for each transmitting node to achieve the hard reliability requirement in MC-IoT applications. However, the RMP-RPL fails to achieve the reliability requirement when the number of source nodes and traffic rate increases. In this paper, a multi-path game theoretic congestion control (GTMP-RPL) approach is proposed on top of the RMP-RPL to reduce congestion at the parent nodes of any child. In case of congestion at any one of the selected k parents, its child node replaces the congested parent by a non-congested parent with minimum rank from the set of remaining nodes. The rank of a node is calculated based on the Data Packet Drop Ratio (DPDR), Expected Transmission Count (ETX), and Node Mobility (NM) of that node. If no non-congested parent is available for that child node, a non-cooperative game is played among siblings to adjust their data transmission rates based on congestion occurrence at the parent, energy spent by the child, and parent connectivity of the child node. The solution obtained for the proposed game, using the Lagrange multiplier and Karush-Kuhn-Tucker (KKT) conditions, is used to reduce congestion at the parent node. The proposed scheme is validated using the cooja simulator in Contiki OS. Simulation results show that GTMP-RPL achieves a 99% packet delivery ratio for at most 50 source nodes with a traffic rate of 30 pkts/min present in a random topology of 101 nodes. In addition, it outperforms the other benchmark schemes by a significant margin to achieve hard reliability.