A mobile ad hoc network (MANET) is an autonomous system of wirelessly interconnected mobile terminals. The interest in such networks stems from their ability to provide temporary wireless connectivity in situations where a fixed infrastructure is lacking or is expensive (or infeasible) to deploy (e.g., disaster relief efforts, battlefields, etc.). Our research is focused on the development of power-controlled medium-access (MAC) and routing protocols for MANETs. Theoretical studies and simulations have demonstrated that transmission power control can provide significant improvements in network throughput (i.e., spatial reuse) and/or reduction in energy consumption. Our main goal is to design fully distributed protocols that enable nodes to coordinate their transmission powers and rates so as to improve the spatial throughout of the network. So far, we have developed several power-controlled MACs for MANETs, including PCDC, POWMAC, and GMAC. PCDC demonstrates the efficacy of cross-layer design, as one aspect of this protocol involves controlling the transmission powers of the route-request (RREQ) packets, hence indirectly impacting the path selection process performed by the routing layer. POWMAC uses a single-channel, single-transceiver architecture (to maintain compatibility with COTS 802.11 hardware). Channel access is performed through a succession of contention slots (together called an access window) during which control packets are exchanged between multiple pairs of terminals. At the end of the access window, concurrent, interference-limited data transmissions can commence in the same vicinity of a receiving terminal at appropriately selected transmission powers and rates. While POWMAC achieves impressive performance gain over the standardized 802.11 approach (i.e., classic CSMA/CA), the heuristic manner by which POWMAC sets the “interference margin” (which is used to accommodate interference from future transmissions) leaves room for improvement. Accordingly, we have recently introduced GMAC (game-theory inspired MAC), in which the determination of transmissions powers is inspired by game-theoretic analysis that is aimed at maximizing a network utility subject to constraints ...[click here for more details]
The main goals of this project are to investigate optimal resource allocation policies for opportunistic and collaborative cognitive radio networks (CRNs), and to use such policies in the design of distributed and adaptive cross-layer protocols for CRNs. Through opportunistic access to the licensed spectrum, CRNs aim at improving spectrum efficiency, hence providing higher spatial reuse, programmable connectivity, and increased network availability. Our research agenda includes analytical formulations that aim at joint optimization of transmission powers, transmission rates, and spectrum in an opportunistic and distributed ad hoc CRN that is co-located with several primary (legacy) radio networks (PRNs). The presence of the PRNs impose frequency-dependent power masks on the transmissions of the CRN. Depending on channel dynamics (i.e., channel’s coherence time relative to the optimization window), we study a deterministic-control formulation for slowly varying channels as well as a stochastic-control formulation for fast varying channels. In both cases, the setup is general enough to allow for multi-channel, multi-path optimization at either the packet level or the session level. The optimized resource allocation strategies are used to develop distributed MAC and routing protocols for opportunistic CRNs under various settings and system constraints ... [click here for more details]
In this project, we aim at designing resource management schemes and adaptive protocols for wireless networks that are capable of directional transmission/reception. In contrast to an isotropic antenna, which transmits the same amount of power is all directions, a directional antenna has preferred direction(s) for transmission and reception; while transmitting, the antenna concentrates the power in certain direction(s), and while receiving the antenna has greater sensitivity for electromagnetic radiation in certain direction(s). Compared with an isotropic antenna, directional antennas have the potential to significantly improve the network throughput and/or reduce the required per-bit energy consumption. For instance, sectoring provided by directional antennas enables a base station to serve more than one cell at a time, thus improving the capacity of a cellular network. Because of these advantages, directional antennas have been adopted in IS-95 and 3G cellular systems. Recently, directional antennas have been suggested for mobile ad hoc networks (MANETs) as well as wireless mesh networks (WiMAX). However, classic MAC and routing protocols in such networks have been designed for omni-directional communications, and extending these protocols to handle directional communications is fraught with challenges. Chief among them is new forms of the hidden-terminal problem (attributed to the asymmetry of the communications), side-lobe interference (due to energy leakage), and transmitter deafness. To address these challenges, we are developing novel, power-controlled MAC and routing protocols for MANETs with directional antennas. Two MAC protocols (called DMAP and LCAP) have been developed and evaluated using simulations and analysis. The latter protocol holds greater promise for practical implementation. Its novelty lies in using an elaborate packet-based power control strategy that is aimed at increasing the channel’s spatial reuse by allowing interference-limited, concurrent directional transmissions to take place in the same vicinity. By employing a separate control channel and by accounting for minor-lobe interference, LCAP alleviates many of the channel access problems that afflict commonly used MAC protocols ...[click here for more details]
MAC and routing protocols are often designed for single-input-single-output (SISO) wireless systems, where each node is equipped with a single antenna for transmission and reception (typically, operated in a half-duplex manner). Significant improvement in performance can be achieved by employing multi-input multi-output (MIMO) techniques, whereby multiple transmit and/or receive antennas are used to provide spatial diversity. MIMO offers three types of gains: array, diversity, and multiplexing. The array gain is achieved either at the transmitter through directional alignment of the transmitted signal or at the receiver by coherently combining independently received copies of the signal. Diversity gain is interpreted as the slope of the average BER curve versus SNR, which is proportional to the number of independent paths. Multiplexing gain is obtained when different signals are transmitted over various transmit antennas for the purpose of increasing the total transmission capacity of the link. In this project, we are investigating the feasibility of adapting the number of transmit/receive antennas (on a per-packet basis) in multi-antenna wireless devices for the purpose of minimizing energy consumption and/or maximizing network capacity. Our initial setup considers networks with 2-antenna nodes, allowing for four possible transmit/receive antenna configurations (also called "antenna modes") per active link: 1-by-1 (SISO), 2- by-1 (MISO), 1-by-2 (SIMO), and 2-by-2 (MIMO). Depending on the type of MIMO gain (diversity vs. multiplexing) and the primary performance objective (energy reduction vs. throughput), we have developed several MIMO-adaptive MACs for MANETs and wireless LANs. These include E-BASIC, MIMO-POWMAC, and CMAC. E-BASIC targets MIMO systems that are capable of providing diversity gain. It allows nodes to distributively adapt their antenna modes, transmission powers, and modulation orders (for multi-rate systems) on a per-packet basis such that the total per-bit energy requirement (transmission plus circuit) is minimized. The protocol was integrated into the design of a power-aware routing (PAR) scheme for MIMO-capable ad hoc networks. In contrast of E-BASIC, MIMO-POWMAC is aimed at multi-antenna systems that offer multiplexing gain...[click here for more details]
Wireless sensor networks (WSNs) are expected to play an important role in a wide range of civilian and military applications, including environment monitoring (e.g., soil and air contaminants), seismic-structure analysis, marine micro-organisms research, military surveillance and reconnaissance, etc. For harsh deployment scenarios, sensors are powered by limited-energy, often non-rechargeable batteries. This makes energy consumption a critical factor in the design of a WSN and calls for efficient and distributed energy management solutions that maximize the operational lifetime of the network. In this project, we are developing such solutions through intelligent topology management, adaptive sensing, clustering, cooperative (virtual) MIMO communications, and robot-assisted gathering and routing of sensed data. Our agenda includes:
Dynamic Activation of Sensors in Location-Unaware WSNs. To improve their reliability, sensors are typically redundantly deployed to account for unexpected failures and improve the fidelity of the received measurements. Redundancy means that some parts of the monitored area are covered by more than one sensor at the same time. While redundancy achieves better reliability, it does not necessarily improve the coverage time, defined as the time until a certain percentage of the area is no longer monitored by any sensor. To prolong coverage time, the network topology can be controlled by selecting a subset of nodes that actively monitor the field and putting the remaining nodes to sleep. We are developing and implementing distributed algorithms that enable GPS-less sensors to determine their active/sleep patterns (including the duration of the active/sleep period) based only on local information gathered from neighboring sensors or from “anchor” nodes (e.g., robots with GPS capability) that roam around and provide routing and location information to sensors. Our work is motivated by many practical scenarios in which sensors are required to operate without line-of-site communications to satellites (hence, the inability to acquire GPS coordinates).
Adaptive
Sensing for Maximized Coverage in WSNs. For certain types of sensors
(e.g., radar sensors), the sensing process is the dominant source of energy
consumption. To prolong the coverage time of the network, sensors can
intelligently adapt their sensing ranges. We are developing distributed
mechanisms for self-adjustment of sensing ranges according to topological
and energy considerations.
Clustering Algorithms for WSNs. When sensors are deployed in large numbers, it is extremely inefficient to operate them as a flat (ad hoc) topology, where each sensor acts as a data source and as a relay to other sensors. In such scenarios, clustered designs are known to provide better network manageability and improved network lifetime. In a clustered WSN, sensors are grouped into dynamically formed clusters, and each cluster is assigned one of its members to act as a cluster head (CH). The CH is responsible for collecting data from the members of its own cluster, fusing (e.g., summarizing, aggregating) the collected data, and then transmitting a summary report to a command-and-control center (CCC). The key questions that we seek to answer here is how to allow sensors to self-cluster with no or little intervention from the CCC, how to dynamically elect and re-elect the CHs, and how to design energy-efficient routing protocols in such a clustered architecture.
Network congestion remains one of the main barriers to the continuing success of the Internet. For WWW users, congestion manifests itself in unacceptably long response times. One possible remedy to the latency problem is to use caching at the client, at the proxy server, or even within the Internet. However, WWW documents are becoming increasingly dynamic (i.e., have short lifetimes), which limits the potential benefit of caching. The performance of a WWW caching system can be dramatically increased by integrating into its design document prefetching (a.k.a., “proactive caching”). Although prefetching reduces the perceived user response time, it also increases network load, which in turn may increase the user response time. Our research in this project is focused on: (1) developing stochastic models for characterizing the behavior of WWW traffic, and (2) using such models in designing, analyzing, and optimizing the performance of hybrid WWW caching/prefetching protocols. For the modeling part, we have used multifractal analysis as a means of characterizing the irregularities in web traffic at multiple time scales ... [click here for more details]
Forthcoming generations of wireless mobile communications systems (3G and beyond) promise revolutionary growth in the achievable data rates of wireless services and in the diversity and flexibility of these services. These systems are expected to support data rates compatible with multimedia applications and to guarantee the QoS requirements of these applications (e.g., packet delay, packet loss rate). Despite recent advances in this area, enabling continuous video streaming over wireless channels is still fraught with challenges. On the one hand, the available bandwidth of a wireless link fluctuates significantly as a result of channel fading and interference. On the other hand, video applications impose tight throughput and jitter requirements, particularly for interactive communications. The situation is further aggravated by the contention-based nature of common wireless access techniques, which gives rise to packet collisions. Collisions can result in packet erasures, whose impact on video quality may extend to several interdependent frames. Our general goal in this project is to provide a unified framework that exploits and jointly optimizes available available adaptation approaches for maintaining continuous video streaming over dynamic wireless links. Our solutions include combinations of source-rate control, adaptive channel coding, and error concealment. These solutions are obtained while accounting for the buffer occupancy at the receiver. Both “archived” and "encoded-on-the-fly” videos are studied ... [click here for more details]
This project aims at the design, analysis, and experimental evaluation of QoS-aware intra- and inter-domain routing solutions, with particular focus on performance and scalability. The goal of QoS routing is to identify a path through a network or series of networks that has sufficient resources to satisfy a set of constraints and, when possible, to optimize the selection of such a path. Typical constraints include bandwidth, maximum delay, reliability, administrative weight, etc. Requesters of constraint-based paths can be end-systems conveying their requirements through protocols such as RSVP, or network administrators attempting to provision paths for traffic engineering purposes. In particular, emerging Internet services such as Differentiated Services and Multi-Protocol Label Switching (MPLS) are likely to both require and justify the need for QoS-based routing solutions. QoS routing represents a radical shift from the traditional connectivity-based routing approach. It brings with it a host of challenges that are being addressed in this project. First, the problem of finding a path subject to QoS constraints is, in general, NP-complete, necessitating the reliance on polynomial-time heuristics and approximate algorithmic solutions. Second, such solutions must often operate under partial knowledge of the network state, which, for scalability purposes, should be aggregated before being disseminated throughout the network. In essence, there is a fine tradeoff between scalability and routing performance that must be accounted for when designing QoS routing solutions. In this project, we are developing efficient solutions for constraint-based intra-domain routing in IP networks, with emphasis on performance and functional enhancements. The proposed solutions include low-complexity path selection algorithms, efficient rerouting techniques, and path restoration mechanisms for reliable routing. We are also developing efficient solutions for constraint-based inter-domain routing, with emphasis on scalability through intelligent reduction of routing information... [click here for more details]
QoS provisioning over wireless networks.
Pseudo self-similarity and chaoticity in network traffic.
Encapsulation mechanisms for real-time and non-real-time packet services over satellite links.
Network support for interactive video-on-demand using time-varying traffic envelopes.
See the web page of the Wireless and Advanced Networking Group for details.