Cross-layer design for quality of service (QoS) in wireless mesh networks (WMNs) has attracted much research interests recently. The network is expected to support various types of applications with different QoS requirements. This can not only be achieved by accurate route selection for QoS, multi-user diversity gain exploitation by scheduling algorithm, but also by efficient connection admission control scheme that can efficiently manage the resources among existing and new flows to improve the overall network performance by considering grade-of-service (GoS) in connection level. Unfortunately, not much work integrates these three layers under the same paradigm with both QoS and GoS provisions. We propose such a unified framework to take advantage the physical channel qualities, multi-user diversity gain in MAC layer, traffic loading and other network conditions to strictly guarantee both QoS and GoS. Studies with different scheduling and routing algorithms, and compared with other admission control schemes, numerical results show that our framework has successfully guarantees various QoS requirements and achieves higher network goodput.

2. MEMBRANE OPNET simulator (MEMBRANE project)

Collaborators: Dr. Thanos Gkelias, Dr. Erwu Liu, Sara Grilli Colombo, Giovanni Paltenghi, and Prof. Kin Leung

Research Contributions: 2 Conference papers and Publised models

Abstract:

MEMBRANE OPNET simulator is a powerful and efficient cross-layer simulation platform developed by Prof. Kin Leung's research group at Imperial College for wireless backhaul mesh networks. In this simulator, we develop, test and compare a set of existing and novel cross-layer protocols from physical layer all the way up to application layer. More specifically, some of the basic MEMBRANE simulator functionalities and models include:  

• Support of various traffic demands with different QoS requirements (such as, interactive video, VoIP calls, FTP, HTTP and Email transfers),   
• Several transport layer TCP protocols (such as, Reno, NewReno and Jersey with and without explicit rate/error/congestion notifications),
• Several IP layer multi-constrained QoS routing algorithms derived in our research group,
• A distributed opportunistic MAC layer (that can support different utility functions) derived in our research group,
• Enhanced physical layer functionalities, such as, adaptive modulation and coding schemes, steering beam directional antennas, and different fading channel representations.

The developed simulator is ideal for protocol verifications and performance analysis on all packet, connection, and network levels.  Compared with existing simulation platform like OPNET WiMax, we believe our simulator is one of the few that considers five protocol layer developments in an integrated simulation environment.

Group OPNET webpage by Harold: http://www.commsp.ee.ic.ac.uk/~chiliu/opnet/

3. Generic Admission Control Methodology for Packet Networks (ITA project)

Collaborators: Dr. Thanos Gkelias and Prof. Kin Leung

Research Contributions: 2 Conference papers

Abstract:

One of the main challenges in packet networks is the time-varying network capacity depending on the number of the underlying data sessions and their quality-of-service (QoS) requirements, which cannot be easily parameterized by a single (or set of) variable(s). However, the accurate network capacity estimation is highly important since it is required for network optimization, admission control (AC) of new data sessions and optimum operation of the served applications.

To address this challenge, a generic AC (GAC) methodology for packet networks is presented, where the subnetwork between any source and destination network router is represented by a black box. Several network-level and traffic-level parameters are used as inputs. A generic mathematical function is then used to map the multiple inputs to a single output parameter, called \emph{QoS performance index}. The bounded value of this index corresponds to the maximum available network resources, defined as \emph{subnetwork capacity}, under the given data traffic volume and requirements. By using Taylor expansion, we are able to predict the impact of new QoS sessions on the subnetwork capacity that is used by a GAC algorithm to maximize the resource utilization while different grade-of-service (GoS) are maintained. The uniqueness of the proposed methodology is its wide applicability in any type of packet networks, wired and/or wireless, independent of the communication protocols or standards used in lower layers.

4. Network Management through Negotiations for QoI in Wireless Sensor Networks (ITA project)

Collaborators: Prof. Kin Leung, Dr. Chatschik Bisdikian, and Dr. Joel Branch

Research Contributions: 3 Conference papers, 3 IBM Res. Tech. Reports

Abstract:

In this work, a novel mission-oriented sensor network architecture for military applications is proposed involving multiple sensing missions with varying quality of information (QoI) requirements. A new concept of mission QoI satisfaction index indicating the degree of satisfaction for any mission in the network is introduced. Furthermore, the 5WH (why, when, where, what, who, how) principle on the operational context of information is extended to capture the changes of QoI satisfaction indexes for mission admission and completion. These allow modeling the whole network as a "black box". With system inputs including the QoI requirements of the existing and newly arriving missions and output the QoI satisfaction index, the new concept of sensor network capacity is introduced and mathematically described. The QoI-centric sensor network capacity is a key element of the proposed architecture and aids controlling of admission of newly arriving missions in accordance with the QoI needs of all (existing and newly admitted missions). Finally, the proposed architecture and its key design parameters are illustrated through an example of a sensor network deployed for detecting the presence of a hazardous, chemical material.

5. Dynamic Control of Data Ferries under Partial Observations (ITA project)

Collaborators: Prof. Kin Leung, Dr. Ananthram Swami, Dr. Ting He, and Dr. Kang-Won Lee

Research Contributions: 1 Conference paper

Abstract:

Controlled mobile helper nodes called data ferries have recently been proposed to bridge communications between disconnected nodes in a delay-tolerant manner. While existing work has explored various trajectory designs for the data ferry by assuming either static nodes or full observations at the data ferry, the problem remains open when the nodes are mobile and the ferry only has partial observations. In this paper, we investigate the problem of dynamic ferry mobility control under limited-range sensing. Assuming the data ferries are capable of sensing node presence within certain range and adjust their movements dynamically, we aim to design control policies that maximize the number of effective contacts.

We provide a comprehensive model of the control framework using Partially Observable Markov Decision Process (POMDP), based on which we study the structure of the optimal policy and propose an efficient heuristic policy which shows significant improvement over the predetermined benchmark. To the best of our knowledge, this is the first data ferry control mechanism that can handle both run-time randomness and incomplete observations.