Silent networking for energy-constrained nodes

Abstract

This paper proposes Silent Networking – a new method to significantly increase the lifetimes of energy-constrained nodes, networks, and mobile user devices. With Silent Networking, each network element powers off some or all of its radio interfaces during their naturally occurring silent periods, i.e., time periods when it does not expect to originate, receive, or relay traffic through those interfaces even if these interfaces are powered on. Each node makes its own autonomous decisions on whether, when, and for how long to power down which of its network interfaces, i.e., without exchanging any explicit or implicit signaling messages with other nodes. Second, network nodes and user devices make such power-down decisions in a way that is transparent to the networking protocols and user applications so that existing and new networking protocols can be used without modifications to support networking as they are used today. Silent networking increases the network and user device lifetimes, adding to energy savings achieved from energy-efficient networking protocols and applications. We present results of laboratory experiments to show that the proposed Silent Networking approach can significantly increase the lifetime of energy-constrained networks and mobile user devices.

Introduction

Impressive progress has been made in developing energy-efficient communication protocols at different layers and in developing dynamic power management techniques for energy-constrained networks such as mobile ad hoc networks and sensor networks [2], [4], [5], [6]. In particular, some methods modify existing communication protocols at different layers to make them more energy efficient. Some define new energy-efficient communication protocols. Other methods dynamically adjust nodes’ radio transmission ranges to optimize overall energy consumption of the network.

This paper proposes an innovative approach that can significantly increase the lifetimes of energy-constrained networks and user devices without modifying existing communication protocols or defining new communication protocols and without changes to user applications. This proposed approach – referred to as Silent Networking – allows network nodes and user devices to power down some or all of their network interfaces when they do not expect to originate, receive, or relay any traffic through these interfaces. The first key idea of Silent Networking is that each network node or user device makes autonomous decisions on whether, when, and for how long it should power off which of its network interfaces, without exchanging any explicit or implicit signaling messages with other nodes. The second key idea of Silent networking is that the network nodes and user devices will make their power-down decisions in a way that will be transparent to the networking protocols and user applications so that existing and new networking protocols can be used without modifications to support networking. In other words, existing and future MAC- and higher layer protocols for reliable packet transport, collision handling, and packet retransmissions can be used without changes. The Silent Networking may increase packet loss slightly. However, the extra packet loss introduced by Silent Networking can be controlled and is similar in nature to the packet loss experienced due to poor radio channel conditions or user mobility. Silent networking increases the network and user device lifetimes in addition to the energy savings achieved from efficient networking protocols (see [6] for examples at various protocol layers) and applications.

Recently, methods have been designed to reduce energy consumption on mobile devices that use high energy-consuming radio technologies such as IEEE 802.11 by powering off these radio interfaces when possible [1], [3]. However, these existing methods fail to adequately address a critical issue: incoming packets may be lost when a radio interface is powered off. Shih et al. [1] proposed to implement a separate low-energy consuming radio on each mobile device in addition to the regular radio interfaces needed to support user applications. This special low-energy radio is used to implement an always-on signaling channel. When the mobile device is not actively in use, the device and its high energy-consuming wireless network interface card are shut off. The low-energy consuming radio will remain powered on to receive incoming traffic and to wake up the mobile device upon receiving incoming traffic. Once awake, the mobile device uses its primary, higher-power and higher-rate channel to transport user traffic. This approach requires special hardware to implement the low energy-consuming signaling channel, which increases the complexity and the costs of the mobile devices. The approach in [3] powers off a network interface on a network node or user device after the interface remains idle for a time threshold called the “think time”. It, however, does not address the issue that incoming packets may be lost when the network interface is powered off. Reference [9] proposes dynamic power management method that let nodes autonomously switch power modes. However, it makes a stringent assumption that incoming packets arrive according to a Poisson process. We have observed from our laboratory experiments that the incoming packets often do not follow Poisson processes. Our method is designed to work even when the incoming packets follow non-Poisson arrival processes. There is also a growing body of work on power-saving methods specific to the IEEE 802.11 protocol, see, e.g., [10]. The Silent Networking method proposed in this paper is independent of networking protocols, and can be used to achieve extra energy savings beyond the energy savings achieved by improving the energy efficiency of networking protocols such as the IEEE 802.11.

Silent Networking is an approach for network nodes and user devices to make completely autonomous decisions on whether, when, and for how long they can power down their network interfaces in a way that will not affect the nodes’ ability to use existing or new networking protocols to maintain a network. Furthermore, nodes will make such decisions in a way that minimizes the probability of losing incoming traffic during the times when the interfaces are powered down. The silent networking method can be implemented completely in software without requiring any extra hardware and can support network nodes and user devices with one or multiple network interfaces.

Section 2 describes the proposed method with its enabling mathematical models. Section 3 presents the results of laboratory experiments we carried out using the proposed method. Section 4 presents approaches for improving the performance of the proposed methods.