Research Thrusts

Wireless Full-duplex Thrust: MAC-Layer Design for Wireless Full-duplex

Existing experimental results show that full-duplex wireless channels can successfully operate only in a narrow interference range. This imposes tight configuration constraints (e.g., works only for narrow channel widths) with limited RF design space in order to leverage the full-duplex advantages. While we anticipate improvements in noise-cancellation circuits and antenna cancellation techniques through continuing research, a realistic scenario is that noise cancellation may not always be achievable considering that WLAN devices share a common license-exempt frequency band with devices operating at high transmission power (e.g., microwave ovens and cordless telephones). The feedback channel can be used to obtain instant knowledge about these operating devices and the impact of their interference on communication parameters such as transmit power, bit-rate, packet size, and packet error. Full-duplex channels can also deliver instant feedback on the channel status to higher layer protocols, which may then adjust their transmission parameters. For example, scalable video transmission rate may be adjusted in response to this feedback. However, not every application benefits from an instant knowledge of the channel status, e.g., unidirectional bursty traffic can operate well on half-duplex channels. Therefore, we see a need for a new MAC design that incorporates full-duplex wireless systems when possible and when needed.

Sensor Network Thrust: Optimal Sensor Deployment and Localization, Communication Aspects

Wireless sensor networks have recently been deployed for data gathering at various ground, sea, and air levels. Such environments are considered harsh for humans to operate in, especially in the presence of high altitude, temperature, and pressure. Due to the costs of placing and maintaining sensor networks, the sensor placement problem has been traditionally solved for a limited number of sensors that often cannot provide adequate coverage of a realistic network. In contrast, the dense number of sensors in certain geographical area causes high interference, in which resulting in generating corrupted data. In addition, existing sensor technology limits the locations for placing sensors, as only a small number of points in an underground network are both accessible and located near a power sources. Therefore, the goal of this research is to further investigate the problem of sensor network deployment and localization in civil systems such as, oil, gas, and water distribution systems. We will also extend our study to investigate the optimal placement of sensor motes in remote geographical areas in which an early event detection and distribution is essential for civil protection and safely (e.g., flood, earthquakes, and volcanoes detections). Optimal placement of sensor nodes is a challenging problem. It has been shown that the problem of optimally deploying sensing infrastructure, i.e., sensor and beacon nodes, and show it is NP-hard. There are heuristic solutions proposed to solve the optimization problem, however, most of them results into unrealistic values. This is mainly due to the fact that many physical communication aspects are ignored.  

In this research we consider localization and coverage of the edges of a graph. Our optimal placement algorithms aim to integrate the sensor node signal, noise, interference, and power aspects into consideration. This is essential to obtain the maximum accurate coverage of sensor nodes and simplify the node localization process. Existing solutions for event detecting and even localization in complex system have been proven costly and imprecise, especially when dealing with large scale distribution systems. In this research, we investigate how mobile sensor networks can be used for optimal event detection in distributed system. Sensor nodes are placed along the edges of the network and detect events and proximity to beacon nodes, with known placement in the network. We formulate the problem of minimizing the cost of the monitoring infrastructure (sensor and beacon nodes), while ensuring a required degree of sensing coverage in an area of interest and a required accuracy in detecting events. We propose algorithms for solving these problems and demonstrate their effectiveness with results obtained from real-life system deployment. The development of such system requires an extensive system analysis and planning where a collaborative control, tracking algorithm, and telecommunication protocols should be carefully coordinated. Furthermore, the system should be stable in the presence of weather, sea, and underground conditions.

Optical Networks Thrust: Failure Localization in All-Optical Mesh Networks

Monitoring and failure localization has long been considered as a critical task in the control and management of backbone carrier networks, and has been a research focus by both industry and academia for achieving a distributed and autonomous environment that can support strict quality of service (QoS) requirements and fast failure restoration. Motivated by the importance of the problem, this research aims to develop a solution plane towards a distributed framework of failure localization in WDM based all-optical mesh networks. The proposed approaches are characterized by launching supervisory lightpaths or optical bursts along a set of pre-defined monitoring routes. Since each monitoring structure is terminated at a monitor, an alarm is issued at the monitor if the supervisor lightpath unexpectedly becomes dark (i.e., loss of light (LoL)), or the launched burst is not received at the expected time instant. With a set of properly designed monitoring routes, the network controller will be able to unambiguously identify the failed links by collecting the alarms issued by the monitors that identified a failure of the corresponding monitoring routes. The proposed research firstly defines the scenarios of shared risk link groups (SRLGs) that are essential to group testing of link status into a combinatorial manner. Then, a novel formulation to the problem is introduced, in which intelligent routing methods for a set of freely-routed monitoring routes under various SRLG scenarios are investigated. Our goal is to efficiently obtain the solution to the monitoring structure routing problem under a suite of constraints, which comprehensively considers the number of monitoring structures, the amount of monitoring resources, and/or the resultant monitoring delay. We expect that the proposed research will yield significant impacts on the research community of optical network fault management, and will create a completely new design paradigm and research dimension to the related research topics. Since the proposed approaches are practical in engineering implementations, we also expect that they can find applications in real-world networks in the near future.

Research Interest

My research interest is in networking and distributed systems, end-to-end QoS, and security. I'm is also interested in managing resources in Grid/Cloud & Autonomous computing, TCP performance over next generation all optical bufferless burst switched networks, wirelss full-duplex, and DTN networks. Summary of my research interest:

• Resource management in high speed optical networks
• Wireless Full-duplex LANs
• TCP/IP over OBS/OPS & wierless networks
• Grid/Cloud Computing
• Internet Security

PhD Thesis

Master's Thesis