ResearchNSFC 62271306: "Design Principle of Large-scale Nonblocking Reconfiguration Optical Add-Drop Multiplexer (ROADM)", 2023/1 - 2026/12
Abstract: Reconfigurable optical add/drop multiplexer (ROADM) has become the core switching device of optical transport network (OTN) because of its flexible optical bypass and add/drop functions. The rapid growth of broadband applications requires a substantial expansion of the port count of ROADMs. The traditional ROADM is a kind of two-stage switching fabric, and the scale expansion is limited by the port count of internal optical devices. Currently, it can only support 30+ ports. The challenge of designing new switching structure is that scaling may result in high losses or sacrifice switching flexibility, both of which are difficult to balance. This project will explore a new modular structure of ROADM that maintains switching flexibility and minimizes loss costs while scaling. Aiming at the problem that it is difficult to balance the loss and the flexibility of optical bypass, this project will study the structural characteristics and expansion principle of ROADM, and propose a matrix-based algebraic method to promote the equivalent transformation of the system structure and achieve low-loss scale expansion. In order to solve the problem that flexible add/drop function imposes a high requirement on switching scale, this project will establish a queuing model to analyze optical add/drop performance to seek a theoretical method for the optimal selection of the number of add/drop ports. The relevant theories and structures proposed in this project can meet the needs of large-scale optical switching in the future, and will provide key technologies for the national broadband network development strategy.
NSFC 61671286: "Modularization principle of passive optical interconnection networks for ultra-large-scale data centers", 2017/1 - 2020/12
Abstract: With the dramatic increase in number of servers in data centers, traditional wiring networks connecting port directly make the cabling complexity of data center networks extremely high, and thus limit the development of data centers. Passive arrayed waveguide grating routers (AWGRs) have the intrinsic attractiveness in the reduction of the cabling complexity, but a single AWGR is not scalable and thus cannot meet the interconnection requirement of Mega data centers. Therefore, we will propose a method to construct modular AWG-based passive interconnection networks in this project, in order to simplify the cabling complexity of Mega data centers. In particular, based on the wavelength routing properties of AWGRs, we will study the wavelength routing properties and the related algebraic structures of AWGR-based multistage interconnection networks, and investigate the condition of wavelength reuse in the network, which will finally yield the modular AWG-based passive interconnection network with the following features: (1) it can simplify the cabling complexity of data center networks, (2) it can preserve the same function and bandwidth as the original data center network, and (3) it exhibits a good scalability since the size of each employed AWGRs and the number of required wavelengths are both small. The results obtained in this project can be applied not only to the simplification of the cabling complexity of Mega data center networks but also the network application scenarios where the cabling becomes a bottleneck.
NSFC 61271215: "Principle of scalable AWG-based optical-electro hybrid crossbars for next-generation green networks", 2013/1 - 2016/12
Abstract: Scalable and power-efficient switching node can reduce the power consumption and the cost of the network during the process of network expansion. Among a number of candidates, optical-electro hybrid crossbar switch, consisting of an arrayed waveguide grating (AWG) and the optical-electrical-optical wavelength converter, has exhibited the superiorities in power consumption, switching speed, and compatibility with the electronic buffer technology. However, the scalability of the AWG is severely restricted by the coherent crosstalk, which restricts the application of large-scale AWG-based optical crossbars. To address this issue, based on the idea of modularity, this project intends to proceed from the routing property of the AWGs and propose to build up large-scale passive wavelength routing systems using a collection of small-scale AWGs, such that the restriction on the scalability of AWG-based optical-electro hybrid crossbar switches can be removed. Also, in order to reduce the cost of the switching systems, this project will build conflict graph models and investigate the optimization problems of the proposed optical-electro hybrid crossbar switches. At last, utilizing the facilities of the state key laboratory, this project will experimentally verify the feasibility of the proposals, and further improve the design of the AWG-based switching structures based on the testing data, such that scalable and power-efficient AWG-based optical-electro hybrid switching fabrics can be achieved. The results of this project will provide key ideas not only for next-generation green Internet but also for large data center network.
NFSC 61201223: "Deflection-based scheduling algorithm for multistage switching networks", 2013/1 - 2015/12
Abstract: Despite the high throughput and low complexity achieved by input scheduling based on Birkhoff-von- Neumann (BvN) decomposition, the performance of the BvN switch becomes less predictable when the input traffic is bursty. In this project, we propose a deflection-compensated BvN (D-BvN) switch architecture to enhance the quasi-static scheduling based on BvN decomposition. D-BvN switches provide capacity guarantee for virtual circuits (VCs) and deflect bursty traffic when overflow occurs. The deflection scheme is devised to offset the excessive buffer requirement of each VC when input traffic is bursty. The design of our conditional deflection mechanism is based on the fact that it is unlikely that the traffic input to VCs is all bursty at the same time; most likely, some starving VCs have spare capacities when some other VCs are in the overflow state. The proposed algorithm makes full use of the spare capacities of those starving VCs to deflect the overflow traffic to other inputs and provide bandwidth for the deflected traffic to re-access the desired VC. The goal of such deflection-compensated mechanism is to support BvN switches to achieve close to 100% throughput of offered load even with bursty input traffic, and reduces the average end-to-end delay and delay jitter.
NSFC 61172065: "Theory and Applications of Multicast Clos Switching Network", 2012/1 - 2015/1
Abstract: The core issue of three-stage Clos networks is nonblocking route assignments of incoming calls for both unicast and multicast traffic. The nonblocking conditions for unicast calls are widely studied, but the well-established bipartite edge coloring approach failed in general multicast cases. In this project, we formulate the route assignment of multicast calls as a vertex coloring problem of conflict graph, whose structure can be uniquely described by its underlying hypergraph. We show that some previously known results can be interpreted by using our approach. Then, we derive a set of new SNB, WSNB, and RNB conditions for multicast calls.
