Cooperative Localization in Wireless Networks

Introduction

The availability of absolute or relative position information is essential in many applications, such as localization services in cellular networks, search-and-rescue operations, asset tracking, blue force tracking, traffic monitoring and autonomous vehicles (Win et al., 2011). Automatic localization of the nodes in a wireless network is a key enabling technology which, mainly because the availability of the node position is highly relevant to the value of the data it collects. With the integration of global positioning system (GPS) into cell phones, in conjunction with other common commercial signals, such as global system for mobile (GSM), WiFi, Bluetooth and LED, the ubiquitous location-aware services with sub-meter accuracy are envisioned in the near future, bringing new opportunities for many wireless applications in commercial, public safety, and military sectors. The wide range of potential applications will also motivate an increasing research interest in localization and tracking technologies for wireless networks.

Location-aware networks generally consist of two kinds of nodes: anchor nodes and agent nodes, where anchors have known positions while agents have unknown positions. Two common examples of anchor-based techniques include the GPS and beacon localization. In GPS, an agent determines its location based on the signals received from a constellation of GPS satellites. However, GPS does not operate well in harsh propagation environments, such as indoors or in urban canyons, since the signals suffer from severe attenuation when across obstacles and non-line-of-sight (NLOS) signals will degrade the localization accuracy severely. Beacon localization, on the other hand, relies on terrestrial anchors, such as WiFi access points or GSM base stations. In areas where network coverage is sparse, e.g., in emergency situations, localization errors can be unacceptably large. Therefore, there is an urgent need for localization systems that can achieve high accuracy in harsh propagation environments with limited infrastructure requirements.

Figure 1 illustrates an example of comparison between traditional and cooperative localization/navigation in a simple network with two mobile agents and four static anchors. Conventionally, each agent determines its position based only on the range measurements to at least three anchors, e.g., Figure 1-(a), which is also known as triangulation. These techniques often fail to localize an agent (e.g., Agent 1) when there are insufficient neighboring anchors within the communication range, or fail to provide satisfactory localization accuracy (e.g., Agent 2) when the range measurements are subject to large errors due to NLOS propagation. However, when the agents cooperate in both spatial and temporal domains, they are able to not only localize themselves with enough line-of-sight (LOS) signals, but also significantly improve the localization accuracy with more valid constraints e.g., Figure 1-(b).

If the agents with unknown locations can make measurements among themselves, such as relative ranges and relative angles, the additional information gained from these measurements between pairs of agents can enhance the accuracy and robustness of the localization system. We call this type of cooperation as spatial cooperation since the relative information in the space is utilized (Patwari et al., 2005). Moreover, if an agent can also incorporate its previous states and then the performance can be further improved. This type of cooperation can be viewed as temporal cooperation since the nodes’ states at different time instants are utilized. The combination of these two types of cooperation in the wireless network enables the joint spatial and temporal cooperation, leading to a new paradigm for real-time localization and navigation in wireless networks, which is referred to as network localization (Win et al., 2011). The joint spatial and temporal cooperation among network also improve the robustness of localization in harsh propagation environments by adding more information.

Figure 1.

A network with two mobile agents (blue rings) and four static anchors (red rings) in two time steps, in which light blue and dark blue rings denote the agents in the previous and current time step. Inter- and intra-node measurements are denoted by red and black arrows, respectively. Anchor 4 provides NLOS signals due to harsh propagation conditions while other three anchors provide LOS signals. (a) Before spatiotemporal cooperation. (b) After spatiotemporal cooperation.