Ms Zhang is a Ph.D. candidate in the NJIT Department of Electrical and Computer Engineering. Her home town is Anhui, China. She received a BS degree in Electrical Engineering from Xi'an Institute of Posts and Telecommunications in 2003, and an MS degree from Shanghai Jiao Tong University in 2006.  Upon graduation, she plans to work in the telecommunications industry.

In the last decade, a dramatic increase in the aggregate information carrying capacity of optical fiber backbone telecommunications networks has been facilitated through the deployment of wavelength-division multiplexing (WDM) technology; which can provide more than 10Gbps per wavelength, on more than a hundred wavelengths per fiber. At the same time, local area networks (LANs) have transitioned from 10 Mbps to 100 Mbps, and are being upgraded to “Gigabit Ethernet” speeds. However, the “access” network, between a home or a business and the backbone network, continues to limit access data rates to 20Mbps, or less, for most residential and small business customers. This “bottleneck” limits the adoption and the growth of bandwidth-intensive applications such as: access to multimedia web sites containing video and high resolution animations, interactive learning, and e-science.

Next generation passive optical networks (NG-PONs), the subject of this research, can eliminate this bottleneck in the “last mile” of the telecommunications infrastructure, between the service provider’s backbone gateway and customer sites.

The major issues we are addressing include:

1. What is the maximum individual-user data rate and what is the aggregate capacity of a given NG-PON architecture.
2. How do we design NG-PON architectures to satisfy a given set of user requirements with the minimum capital expenditure on terminal equipment and the optical distribution network (ODN).
3. How do we tackle the resource allocation issue in NG-PONs, which usually involves both time allocation and wavelength allocation of network assets to current network users?

For issue 1, we propose to: abstract the data transmission process in NG-PONs, using a tiered directed graph; to search for special relationships between vertices in the tiered graph; and to perform a subsequent derivation of the achievable ranges of individual user data rates and aggregate capacities of those NG-PONs. 

For issue 2, the major capital expenditures in building a NG-PON are: the cost of the ODN, including the right-of-way acquisition and the installation cost; the optical line terminals (OLTs); and the optical network units (ONUs). To minimize the ODN cost, we employ cascaded Arrayed Waveguide Gratings (AWGs), and propose heuristic* algorithms to optimize the tradeoff between AWG cost and optical fiber cost. (*This is an NP-hard problem). To minimize the ONU and OLT cost, we provide guidelines for reducing the number of lasers and for decreasing the use of expensive full-range-tunable lasers, without compromising the aggregate network capacity. 

For issue 3, we investigate the impact of lasers’ tuning ranges and tuning speeds on the system performance. Specifically, we apply bipartite graph matching to the resource allocation problem, for the scenario in which lasers are limited-range-tunable. We tailor our earlier works on multi-processor scheduling for the resource allocation problem with respect to the scenario in which lasers need some time to change their wavelengths. Furthermore, we are also investigating the “max-min fair resource allocation” issue for diversified applications in NG-PONs; which can be abstracted as a multi-objective optimization problem.