Adaptive Communications Research Laboratories

Research on Wireless Ad Hoc Networks




KOMIYAMA Bokuji, Director


1. Introduction
　　The growth in mobile communications has been explosive. In addition to speech services, mobile data services have also become widespread. With the introduction of third generation (3G) systems in 2001 in Japan, the growth of mobile internet services has also accelerated. Recently, however, new demands have appeared, like inter-car communication or mobile peer-to-peer communication, which conventional cellular systems with centralized control cannot effectively serve. In order to satisfy these demands, there is growing interest in ad hoc mobile networks.
　　Ad hoc networks are networks with decentralized control, which consist of mobile nodes equipped with a wireless transceiver. Key features of ad hoc networks are to establish communication, anytime and anywhere without the aid of a central infrastructure like base stations or access points, and to extend the communication range by use of multiple hops. Fig. 1 shows the expected performance of wireless ad hoc networks compared to other wireless networks. Higher data rates than conventional large-scale cellular systems can be expected, although they are limited to shorter range communication, and there is no communication charge, in principle.
　　Typical applications of ad hoc networks are networks in disaster areas, exhibitions, and campuses. They might also be appropriate for home networks, and links between factory robots and sensor networks. Most research work on ad hoc networks has been performed by simulation, because it is not feasible to construct small-size test beds for experiments in the laboratory.
　　In January 2002, we launched our new project, supported by TAO, with a time scale of three years and three months. The purpose of the project is to develop key technologies for wireless ad hoc networks from physical layers to upper layers. We emphasize experimental studies and aim at examining network performance in a real world environment with a laboratory-size test bed. This article gives a brief overview of the project.

2. Research areas
　　A network environment of ad hoc networks is more dynamic than traditional wireless networks, because some or all terminals (nodes) are mobile and terminals can suddenly disappear from the network, as shown in Fig. 2. The main technical issues of a wireless ad hoc network are the dynamic topology of the network, rapid changes of link quality due to multipath effects in the wireless environment, and the distributed operation of the network. Further, compared to cellular systems in which one-hop wireless links are used, an ad hoc network requires multihop wireless links, which leads to more complex behavior. To address these problems, we are working on the four research topics shown in Fig. 2. Details of each topic are described below.

Network design and control
　　In this sub-theme, the three issues shown in Fig. 3 are being studied. The first issue is MAC (Media Access Control) and routing protocols. The dependence of protocol characteristics on parameters such as the number, density and mobility of nodes is studied theoretically and simulated. In addition, an in-house test bed with low signal power will be constructed, and measured field data will be fed back to the design of optimal MAC and routing protocols. This test bed will enable the examination of the combined characteristics of a MAC protocol and a routing protocol. Since the beam directivity of an antenna installed in a mobile terminal has remarkable potential for increasing performance within ad hoc networks, we are also investigating the effectiveness of directional antennas. The interaction between layers for omni-directional antennas will be investigated first, and then, based on these results, the use of directional antennas or adaptive antennas will be considered.
　　The second issue is the study of adaptive QoS (Quality of Service). Since the expression of QoS is different from layer to layer, and the degree of QoS should eventually be judged by end-users, a layered QoS model should be introduced with a hierarchical structure from user to system. We propose a framework to integrate the adaptive QoS functional structure, and study its theoretical properties and mechanisms with a layered structure. As an important element technology on each layer, we will study the adaptive security function and try to improve TCP (Transmission Control Protocol) performance for the wireless and multi-hop environment. We are also studying resource management adaptive to the characteristics of multimedia in distributed multimedia computing applications.
　　The third issue is the realization of multimedia applications, such as a Voice over Internet Protocol (VoIP) application, in ad hoc networks. We are also developing a Contiguous Communication Networking over Ubiquitous Transmission (CoCoNUT) system, which establishes self-organizing connections among nodes that physically approach one another. To produce attractive services over ad hoc networks, we are developing applications for short distance wireless networks, for utilization by the local community.

Network theory
　　This sub-theme evaluates the performance of MAC and routing protocols in wireless ad hoc networks to elucidate the general rules underpinning them. We intend to clarify limitations on the network size, as measured by the transmission distance (allowable number of hops) and allowable number of terminals. Network performance measures like transmission speed and throughput, and the allowable mobility of the terminals will also be evaluated. In addition to these functions, we will also examine optimized allocation of resources, such as bandwidth. These studies will enable us to determine significant performance measures and guidelines for system design, and to see more clearly from a technological point of view the application possibilities of the networks.
　　In these studies we will give special attention to the directivity of the antennas attached to the terminals, as they are expected to provide enhanced system performance. A qualitative image of network performance is shown in Fig. 4. Theoretical analysis and computer simulations are used in our investigations.
　　Also, we will examine the application of wireless ad hoc MAC/routing technology to intelligent transport systems (ITS). We are examining link establishment and information hopping in inter-vehicle communication for collision prevention at intersections, as well as rear-end collision prevention and co-operative driving on expressways.
　　In this sub-theme, ad hoc networks are studied from a different point of view, that is, from the viewpoint of users. Through the analysis of questionnaires and market surveys, we are examining the needs for ad hoc networks, and mechanisms for the emergence of new market needs. A framework, including concepts and measures, is also being developed for network evaluation. Apart from requirements for network function and performance, what services are preferable will also be addressed. We will determine what types of ad hoc networking are preferred by users.

Personal wireless links
　　A wireless ad hoc network consists of only battery-operated user terminals, and how to efficiently use the limited radio resources, frequencies and power, is a key issue. Our approach is to utilize a space division multiple access technique (multiple access with antenna radiation pattern control). In particular, we aim at developing an adaptive antenna for mobile user terminals.
　　At ATR, an electronically steerable passive array radiator (ESPAR) antenna has been developed for mobile user terminals. The ESPAR antenna has only a single active radiator connected to the RF (Radio Frequency) receiver or transmitter, as shown in Fig. 5. This active radiator (#0) is surrounded by passive radiators (#1-#6) loaded with variable reactors (varactor diodes). Adjusting the reactance value (x1-x6) with the bias voltages on the varactors can change the radiation pattern of the antenna. Two main problems need to be solved for practical use. One is how to miniaturize the hardware such that the antenna can be attached to user terminals. The other is to develop adaptive control algorithms that make the ESPAR antenna adaptively steer its beam toward a desired signal and form its nulls in the direction of interference signals. Most of the algorithms developed for conventional adaptive arrays are useless for the ESPAR antenna due to its simple architecture. Since a well-known training-based adaptive control algorithm would require significant overhead in the form of transmission of training sequences at regular intervals, our goal is to develop blind adaptive algorithms for the ESPAR antenna, where the receiver does not require a training sequence.
　　In addition to measuring electrical characteristics of ESPAR antennas in a radio anechoic chamber, and verifying their performance with respect to radio wave interference suppression, we will evaluate the performance of an ad hoc network equipped with ESPAR antennas in the test bed.
　　Conventionally, even for measurements of small antennas like ESPARs, a large and expensive room-size radio anechoic chamber has been required. In order to reduce the size of such chambers, we are working on techniques for antenna measurement in a compact radio anechoic box. For this purpose, we will apply advanced microwave photonic techniques, where electro-optic or magneto-optic probes are used to measure electromagnetic fields very near antenna elements, without disturbing their original fields.

Microdevices
　　Optical wireless links will be used complementarily with radio wireless links in future ad hoc networks. In this sub-theme, we are conducting research on the key "microdevices" for optical wireless transceiver modules. The microdevices are one-chip optoelectronic devices and have functions suitable for optical wireless links, in particular, powerful beam control and steering functions. The subjects to be investigated are the micro-machine technology called MEMS (micro-electromechanical systems) as well as optoelectronic devices.
　　Light emitting devices, photo detectors and micro-mirrors are the target microdevices to be developed, for which we chose GaAs (gallium arsenide) compound semiconductor as the basic material. Light emitting and detecting devices suitable for integration and implementation in array structures, such as vertical cavity surface emitting lasers (VCSELs), light emitting diodes (LEDs), and photo detectors (PDs) will be experimentally fabricated. Passive microdevices such as micro-mirrors, structures effective for reflecting and controlling light beams, will also be designed. The image in Fig. 6 shows an example of the target microdevices.
　　For the fabrication of these devices, we make use of two techniques invented at ATR: Lateral p-n Junction to fabricate a p-n junction perpendicular to a substrate with one step epi-layer growth, and Micro-Origami to fabricate a microscale three-dimensional structure by simple process steps similar to "origami" (the Japanese art of paper folding).
　　Lateral p-n junction has potential for realizing highly efficient optoelectronic devices with simple structures, and Micro-Origami is a viable fundamental technique for developing MEMS devices with complex three-dimensional structures. 　　In addition, we are investigating driving mechanisms for dynamic control of optical beams using electrostatic force or electro-thermal expansion. Sub-micron processes are another research task essential for the development of devices using diffractive optics, which are needed to attain compactness and high efficiency of optical modules.
　　In parallel with the themes mentioned above, we are also conducting basic research on GaAs-based VCSEL, LED, and PD with long wavelengths (from 1.3 to 1.6_m), which is essential for realizing the eye-safety and low-price requirements necessary for microdevices used in optical wireless links.

3. Conclusion
　　In previous work, funded by the Japan Key Technology Center, the Adaptive Communications Research Laboratories were concerned with adaptation to network and client variability, and adaptation to radio wave environments. In this, our new TAO-funded research project, we are intensifying our focus on ad hoc network-related research topics. We are seeking to comprehensively develop fundamental technologies, from antennas to networking protocols. With the development of these technologies, we will be able to provide the basis for technical guidelines on the practical use of ad hoc networks.