Modeling of a Humanoid and Multi-agent System EE382C: Embedded Software Systems Literature Survey Yuklai Suen Dept. of Electrical and Computer Engineering The University of Texas at Austin Abstract. This survey briefly describes the project to implement an infrastructure for multi-agent systems (MAS) using humanoid robots as agents (HMAS). In order to construct a wireless network that facilitates the design of the infrastructure, this paper reviews some current technologies and researches in wireless networks. The research focused on the following layers of the International Standards Organization (ISO) layered model - Open Systems Interconnect (OSI). There are different modulation proposals of Ultra-wide Band (UWB) technology at the physical layer (PHY). The media access control (MAC) protocols are discussed in their facilities of low power dissipation, mobility management, and assurance of the Quality of Service (QoS). The survey also studies different network designs and evaluates their applicability in HMAS project. The choices for the network designs include cellular network, ad hoc network, and sensor network.I. INTRODUCTION MAS is a sub-field of AI that aims to provide both the principles for construction of complex systems involving multiple agents and the mechanisms for coordination of independent agents’ behaviors [16]. Generally, an agent is capable of detecting the environment, deciding actions based on the detection, and executing the actions. This projects aims to develop a flexible infrastructure for realizing multi-agent systems for educational and research purposes. The infrastructure has a base-station to handle heavy duty computations and a number of humanoids to act as agents. Recently, studies in multi-agent systems have explored areas such as robot soccer (RoboCup) and mobile sensor networks. Realization of these MAS systems through HMAS will be valuable to these researchers. The following sections will describe some prior works in developing a wireless network from the physical layer to the network layer of the OSI model. Section II introduces the specification of the network designs in HMAS and the challenges of the project. Section III, IV, and V introduce the recent researches in UWB, different MAC protocols, and different network designs for this application. Section VI briefly describes the future implementation of this project and summarizes this survey. II. HUMANOID MULTI-AGENT SYSTEMS The infrastructure needs a high-quality wireless communication network that is mobile, cheap, and fast. UWB has become a very good candidate for this purpose. However, the implementation faces a few challenges. First of all, transceivers of UWB are very expensive because the lack of mass production. Secondly, the IEEE 802.15.3a protocol to be used with UWB is not yet standardized [7].Moreover, simulation tools with UWB are not common. Further, further researches reviewed that there are no current protocol designs at the MAC layer that are directly applicable to our project. The proposed solutions to these problems would be addressed in Section VI. III. PHYSICAL LAYER A. History of UWB The Federal Communications Commission (FCC) defined a UWB device to be “any device where the fractional bandwidth is greater than 0.25 or occupies 1.5 GHz or more of the spectrum” [7]. FCC also allocated 7.5GHz of spectrum for unlicensed use of UWB devices in the 3.1 to 10.6 GHz frequency band. Although its applications are new to many people, the concept originated from the 19th century. In 1978, C. L. Bennett discussed the application of short-pulse radio signals in time-domain electromagnetism [4]. IEEE 802.15.3a Task Group (TG3) adopted UWB as the physical layer of the Wireless Personal Area Network (WPAN) [8]. Different modulations are proposed for UWB over the last couple of years [14], [15], and [2]. After a long time of debating, the IEEE 802.15.3a task group (TG3) approved a dual-PHY approach with a single-band, direct-sequence-CDMA proposed by Xtremespectrum and Motorola, and multi-band-OFDM proposed by the Texas Instruments/Intel-led Multiband-OFDM Alliance (MBOA) [16]. B. Why UWB? We select UWB to be the physical layer of our network is because it fulfills our design constraints. UWB has a large frequency spectrum and large bandwidth. It is able to transmit a large amount of data through short ranges within 10 meters. The way that impulse trains of UWB signalpropagates allows simple transceiver design and hence the hardware cost is [14]. The only drawback for UWB is its performance decline over incremental transmission range. IV. MEDIA ACCESS CONTROL A. Definition MAC is a sub-layer in Data Link (DL) layer of the OSI model. It defines the protocol for packet transmission and it acts as a interface between the network and the physical layer. [11] studies different MAC protocols and evaluates their performance on channel acquisition time using UWB as the physical layer. The following sub-section presents a literature in the MAC level of networking for sensor network. Sensor network will be discussed in Section V. B. Protocols for Self-Organization of a Wireless Sensor Network Sohrabi et al presents a set of algorithms to instrument a sensor network that had a strict constraint on low power consumption [12]. They assumed that most of the nodes in the network were stationary once after deployment, and that the number of mobile nodes was small. First, they described how they combined the neighbor discovery and channel assignment phases, which were separated phases in many algorithms, to construct a stationary network. In their algorithm, each node would randomly pick a frequency to avoid collision in communication. They adopted the time-division multiple access (TDMA) approach that required each node to reserve time slots for neighbors. These slots are brought up periodically during which the node could communicate with one of its neighbors. These network propagated the slot allocation schedules until all the nodes were connected to enable multihop communications.Secondly, they presented the Eavesdrop-And-Register (EAR) algorithm to address the mobile node issues. They again assumed that the stationary neighbors of the mobile agents would broadcast invitation messages to all the surrounding nodes. When the mobile node received an invitation, it decides whether to disconnect from a node or to connect to a node based on geographic, energy, or transmission