مسیریابی جغرافیایی برای شبکه های با اتلاف و کم قدرت اینترنت اشیا
Abstract
1- Introduction
2- GROLL Overview
3- GROLL Design
4- GROLL Routing Engine
5- Related Work
6- Conclusions
References
Abstract
Routing in low-power lossy networks (LLNs) remains an important research problem for Internet of Things (IoT), despite the recent standardization in RFC 6550 of RPL as an IPv6 Routing Protocol [1]. Recent deployment experiences [2] have emphasized the challenges faced by a complex design in real world deployments. One of the key challenges is versatile routing protocol design that can handle efficiently and seamlessly various combinations of routing primitives, namely unicast, multicast, and convergecast. Although RPL supports these traffic types, due to its design principle oriented around data collection, its unicast and multicast functionalities are limited. In this article, we present the design and system implementation of the first Geographic Routing framework for Low Power and Lossy Networks (GROLL). GROLL’s routing engine effectively integrates routing for unicast, multicast, and convergecast and takes into account Complex Network Topologies (CNT), i.e., network holes and cuts. GROLL detects CNTs and abstracts the boundary information of detected CNTs with significantly reduced memory size. The proposed unified routing framework was implemented on real hardware and it was evaluated extensively on a testbed of 42 TelosB motes.
Introduction
RPL was adopted by IETF as the standard IoT routing protocol for low power and lossy networks [1]. Recent deployment experiences [2] have emphasized the complexities faced by RPL routing in real world, and pointed out some inefficiencies, especially in terms of routing path stretch. RPL is designed to support unicast, multicast and convergecast, but its design makes limiting assumptions: the predominant traffic is from nodes to the base station, followed, on rare occasions by traffic from the base station to the nodes and the traffic among nodes, while supported, is extremely rare. These assumptions may not hold well in some deployments, such as wireless sensor deployment for disaster response. On the other hand, geographic routing is well suited for LLNs because it is simple and highly scalable; and more importantly location information of devices, sensors, and machines are often readily available in IoT applications. Furthermore, different routing primitives such as 1-1 (unicast), 1-n (multicast), and n-1 (convergecast) can be easily supported based on geographic routing. Unfortunately, to the best of our knowledge, there is no unified location-based routing framework that integrates location-based routing protocols developed for different routing primitives. However, the needs for an unified routing framework has ever increased recently. In recent IoT applications [3], nodes do not just report data to the base station. For example, peer-to-peer or multicast data communications have become essential routing traffic patterns for in-network processing, data aggregation, and feedback control [3].