Efficient Communication in Named Data Networks for Internet of Things

Abstract

Integrating low-cost sensing and actuating devices into the Internet has propelled the evolution of the Internet of Things (IoT), characterized by many small smart devices, or "things," equipped with sensing, communication, and computing capabilities. However, limited processing power, memory, and energy resources often constrain these devices. Additionally, they frequently operate within Low-power and Lossy Networks (LLNs), leveraging IEEE 802.15.4 communication technologies to facilitate data exchange within the IoT ecosystem. Communication in LLNs poses significant challenges, particularly ensuring reliability and adapting to dynamic network topologies resulting from lossy links and device mobility inherent in IoT applications. Named Data Networking (NDN) has emerged as a promising alternative to IP for addressing the communication needs of IoT applications. NDN's data-centric model aligns well with IoT requirements, facilitating user mobility and data sharing through features like caching, naming, and stateful forwarding. However, when deployed over LLNs, nodes in NDN rely on the broadcast nature of the shared wireless medium to forward interest packets to their neighboring nodes, which often results in the broadcast storm problem. This study is among the first to explore the effectiveness of probabilistic techniques for interest forwarding in NDN over LLNs to mitigate the adverse effects of the broadcast storm problem while adhering to LLN constraints. To this end, three distinct probabilistic forwarding strategies—Probabilistic Forwarding (PF), GOSSIP, and Distance-based Probabilistic Interest Forwarding (DPIF)—are introduced. While PF employs probabilities for forwarding control, GOSSIP augments PF through flooding and duplicate control. On the other hand, DPIF augments GOSSIP through propagation control. The rationale for proposing these strategies is to assess the efficacy of combining probabilities with different control mechanisms for interest forwarding in NDN over LLNs, thereby providing a comprehensive understanding of their capabilities and limitations under different operating conditions.Extensive simulations are conducted to undertake the first comprehensive evaluation of interest forwarding strategies in NDN over LLNs. This evaluation encompasses a performance analysis of PF, GOSSIP, and DPIF alongside well-established existing strategies, including Blind Flooding (BF), Deferred Blind Flooding (DBF), Learning-based Adaptive Forwarding Strategy (LAFS), and Provider Aware Forwarding (PAF), across a range of scenarios. Extensive simulation results are collected to quantify and analyze the performance advantages of probabilistic strategies for interest forwarding in NDN over LLNs in terms of packet retransmissions, retrieval latency, success rates, and energy consumption.