DPM-RPL: Routing-Based Distributed Dynamic Power Management for NVP IoT Networks

The RPL routing protocol suffers from inefficient energy consumption, often leading to energy holes in IoT networks. Although RDC-based routing protocols try to reduce energy consumption by accounting for MAC radio duty-cycling, their gains are limited to transceiver-level optimization. These approaches neglect system-level power management, and do not push IoT devices into deep sleep state, keeping the processor, memory, and peripherals active because nodes are assumed unable to retain routing state when powered down. Achieving deeper energy savings requires preserving routing state across power cycles, which is only possible with Non-Volatile Memory (NVM). Building on this observation, this paper proposes DPM-RPL, a distributed routing-based Dynamic Power Management protocol for Non-Volatile Processor (NVP) IoT networks. By embedding system-level power management directly into the network layer, DPM-RPL enables nodes to safely enter deep sleep states while retaining essential routing and operational states. DPM-RPL revolutionizes energy savings by allowing the device’s always-on components to be fully powered down. To provide routing state retention and node synchronization across power cycles, DPM-RPL introduces a novel Deep Sleep Cycle (DSC) metric, enabling nodes to estimate the sleep and active periods of candidate parents and select those with the longest overlapping active durations. This metric improves synchronization between nodes, reducing missed channel checks, strobe-induced energy drain, retransmissions, network disconnections, and control over-head, thereby improving energy efficiency, Packet Delivery Ratio (PDR), and network stability. Additionally, when calculating DSC, DPM-RPL uses Exponential Moving Average (EMA) forecasting to estimate the time between application events, allowing it to adjust sleep and wake periods to match activity patterns. This approach minimizes energy waste due to peripheral reinitializations and excessive cold start and ramp-up phases of critical hardware components such as phase-locked loops, providing energy-efficiency, communication reliability, and reduced latency. DPM-RPL primarily targets dynamic energy reduction, while the incorporation of NVM further decreases energy consumption by minimizing leakage during deep sleep. Extensive simulations demonstrate that DPM-RPL reduces average energy consumption by up to 2.4× (140%) compared to the state of the art, while improving PDR by 37%, reducing latency by more than threefold, and decreasing retransmissions by 57%.