OPNET Technologies
7255 Woodmont Avenue
Bethesda, MD 20814
Tel: 240-497-3000
Fax: 240-497-3001
E-mail: university@opnet.com
Web: www.opnet.com

OPNET is a registered
trademark of OPNET Technologies
© 2004 OPNET Technologies

University: University of Washington
Name of sponsoring Professor: Sumit Roy
Department: Electrical Engineering

1. Wireless Mesh Networks: PHY/MAC based optimization
   
   An important challenge in wireless networking is  designing an appropriate
   Broadband Wireless Access Network (BWAN) to provide  client devices with
   access to a backbone network (the Internet). There exist various
   possibilities for BWAN architectures based on  emergent next generation
   high-rate standards (IEEE groups 802.11, 802.15 and  802.16) that embody
   different link (PHY) and data-link (MAC) layer designs  aimed at different
   applications and usage models.
   
    Clearly, one of the primary attributes of any  intermediate transport BWAN
   network must be network scalability - i.e., it's throughput must scale with
   a) increasing peak data rates per client and b)  larger number of clients.
   Traditional cellular architecture scales by  reducing cell-size via PHY/MAC
   innovations - while this increases aggregate  throughput, it incurs greatly
   increased infrastructure cost since an  expensive base station is required
   within each cell whose numbers (and hence  cost) increase geometrically.
   Accordingly, a MESH network wherein the nodes  act as routers and forward
   packets from other nodes is seen as an attractive  alternative. In such a
   network, only a fraction of the nodes are gateways  (i.e. are connected to
   the backbone network) while a  majority are only (cheap) routers; messages
   are routed end-to-end via intermediate router nodes without requiring a
   direct connection from source  to destination. Such an architecture has
   several compelling advantages: other  than it's potential for
   cost-effective network scaling, they are robust since  they offer multiple
   paths between source-destination pairs that allow  avoidance of
   intermediate congestion points .  However, several key aspects of this
   architecture and it's implication on  network level performance is not
   sufficiently understood at this time - in  particular, there remains
   lingering doubts regarding it's scalability.
   
   We plan to
   
   a) Evaluate available PHY/MACs  for their suitability with respect to the
   MESH architecture, and suggest enhancements to any given PHY/MAC for
   optimal  MESH performance.
   b) Identify and investigate the interactions and inter-relations between
   (i) Traffic Patterns, (ii) MAC and Routing protocols and  Mesh architecture
   and their impact on aggregate network metrics.
   
   
2. Performance of 802.11e EDCA in multi-hop environments
   
   IEEE 802.11e Medium Access Control (MAC) is an emerging supplementary
   standard to support quality of service (QoS) for IEEE 802.11. The Enhanced
   Distributed Channel Access (EDCA) in 802.11e is an enhanced version of
   legacy DCF, and provides contention-based differentiated channel access for
   frames with different priorities. At present, there is already a lot of
   research work done on the performance analysis and simulation of EDCA in
   one-hop WLAN.
   
   We plan to explore the performance issues of EDCA in a multi-hop
   environment. The proposed research involves  the following aspects, each of
   which will involve simulations in part:
   
   a) Evaluate EDCA performance in terms of throughput, delay, and jitter in
   the multi-hop environment.
   
   b) Compare EDCA performance both in multi-hop and one-hop environments and
   identify factors that may cause performance degradation.
   
   c) Provide possible mechanisms to optimize EDCA performance in the
   multi-hop environment.
   
   
3. Capacity and performance of 802.11 networks under asymmetric traffic
   patterns
   
   In an 802.11 network operating in infrastructure mode, the access point
   (AP) will typically have considerably more traffic to transmit than any
   individual client. Two extreme cases where this is apparent are
   a) All traffic originates and terminates at clients in the same cell, and
   therefore the AP acts as a relay for all the traffic in the cell, and
   b) All traffic originates external to the cell and therefore the AP
   transmits all the traffic in the cell (with the exception of feedback
   traffic from the clients, which will be negligible in comparison).
   
   Therefore it is of interest to study a network in which the traffic pattern
   is asymmetric, i.e. one client has considerably more data to transmit than
   the others. However, studies in the literature have typically considered
   symmetric traffic patterns, where all nodes are identical. We will use
   Opnet simulations to develop an understanding of the impact of asymmetries
   on network performance measures such as throughput and delay. We will then
   consider enhancements (such as provision of priorities via 802.11e) in
   order to improve the performance of the network under these traffic
   considerations.