Zengji Liu

Xidian University, Ch’ang-an, Shaanxi, China

Are you Zengji Liu?

Claim your profile

Publications (36)22.52 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: More and more attention is focused on direct interconnection networks when designing the switching fabrics in the terabit routers. Various congestion control mechanisms are proposed for multi-computer systems, which also rely on direct interconnection networks between processors to support the messages passing mechanism. But it remains unknown whether they are suitable for fabrics in the terabit routers. Based on the requirements of terabit class routers, we made analysis and simulations on various congestion control mechanisms, such as U-channel, INC, self-tune and GRCC. The results show that congestion control mechanisms eliminate performance degradation for loads beyond saturation, keeping adequate levels of throughput at high loads. The local mechanisms can achieve similar performance with the global ones. However, as we pointed in the paper, they all have limitations when used in switching fabrics of the terabit routers.
    Communications and Information Technology, 2005. ISCIT 2005. IEEE International Symposium on; 11/2005
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: A methodology is proposed to handle problem that under equiproble address of packet traffic at the input port, Generalized Shuffle-Exchange Network (GSEN) routes traffic unevenly because of the unbalanced routing tags. The idea is to use routing tag according to probability, which can be evaluated by using Moore-Penrose inverse in matrix analysis. An instance is used to illustrate the idea, and the simulation is done to show the improvement in performance issues.
    Journal of Electronics (China) 06/2005; 22(4):345-350.
  • Yan Shi, Zengji Liu, Zhiliang Qiu, Min Shen
    [Show abstract] [Hide abstract]
    ABSTRACT: Satisfying critical QoS requirements in next generation networks poses major challenges, due to its intrinsic complexity of network resource allocation. This paper considers the problem of load balance based bandwidth allocation for delay sensitive services. For satisfying deterministic end-to-end delay requirement, a simple and efficient algorithm for path level optimal bandwidth allocation, path level equal ratio allocation algorithm (P-ERA), is developed first. Based on P-ERA, another algorithm for network level optimized bandwidth allocation, network level equal ratio allocation algorithm (N-ERA), is also presented for much more complicated network circumstances. N-ERA algorithm features appropriate route selection and balanced bandwidth allocation, and can adaptively avoid, or at least delay, the emergence of network bottleneck when a network is heavily loaded. Extensive simulations indicate that N-ERA algorithm can make full use of network bandwidth and admit more services, even if they are delay critical, than other ones dealing with the same problem. And the less computation complexity of N-ERA algorithm makes it of great application value.
    Advanced Information Networking and Applications, 2005. AINA 2005. 19th International Conference on; 04/2005
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To design a Banyan network with an arbitrary even-sized port number, the PN2I network is proposed. The PN2I network can be divided into two classes: the complete and incomplete versions. A simple routing algorithm is given, but in the incomplete PN2I networks, this routing algorithm fails to make the traffic in links even, which deteriorates the performance badly. Thus a new routing algorithm is proposed, which makes the incomplete PN2I networks behave almost the same as the PN2I networks with respect to the performance issues.
    Journal of Electronics (China) 04/2005; 22(5):470-477.
  • Yang Su, Zhanqi Xu, Zengji Liu
    [Show abstract] [Hide abstract]
    ABSTRACT: An extended layered graph of MPLS over WDM mesh networks is proposed in this paper, in which the label switch path (LSP) with various wavelengths and the limitation of optical transceivers at a routing node are both involved. Label switch paths are classified into different priorities according to each quality of service. The corresponding routing algorithm, differentiating integrated routing algorithm (DIRA), is proposed and studied. The quality of service (QoS) of a label switch path and the optimization of network resources utilization are taken into account comprehensively in DIRA. A comparison of DIRA with the representative optical routing algorithms via simulation shows that it can reduce the blocking probability of delay-constraint LSP and improve the network throughput.
    Proc SPIE 01/2005;
  • [Show abstract] [Hide abstract]
    ABSTRACT: More and more attention is focused on direct interconnection networks when designing the switching fabrics in the terabit routers. Various switching mechanisms are proposed for multi-computer systems, which also rely on direct interconnection networks between processors to support the messages passing mechanism. But it remains unknown which one is more suitable for fabrics in the terabit routers. Based on the requirements of terabit class routers we made analysis and simulations on various switching mechanisms, such as store and forward, wormhole switching, virtual cut through switching and pipelined circuit switching. The results show that virtual cut through exhibits superior performance characteristics over other switching mechanisms under various conditions. Simulations of the performances of virtual cut through shows that larger buffer, longer flit and more virtual channels help to sustain higher throughput at the cost of increasing latency.
    Networking - ICN 2005, 4th International Conference on Networking, ReunionIsland, France, April 17-21, 2005, Proceedings, Part I; 01/2005
  • [Show abstract] [Hide abstract]
    ABSTRACT: Direct interconnection network has been widely used in supercomputer systems. Recently, it is considered to be used to build terabit router by the industry. This paper presents a distributed and scalable switching fabric based on a new direct interconnection network. It is a scalable topology and can be expanded in two ways easily, thus minimizing the initial investment of service providers. Its distributed control can offer low hardware complexity. Virtual cut through switching is used to achieve high throughput and low latency. The quality of service is guaranteed by introducing virtual channels based on the concept of DiffServ. The fault tolerant and load-balanced routing algorithm can offer deadlock and livelock freedom. It helps the network continue to work even with faulty parts existed. Finally, the performance of the proposed switching fabric is evaluated under various conditions. The results show it can outperform its counterpart in latency and throughput. It achieves terabit throughput with average latency of 1 us or so.
    Advances in Computer Systems Architecture, 10th Asia-Pacific Conference, ACSAC 2005, Singapore, October 24-26, 2005, Proceedings; 01/2005
  • [Show abstract] [Hide abstract]
    ABSTRACT: Research is carried on a novel switching fabric-XD (cross and direct) network, which is the core of the terabit router. XD networks have many advantages such as simple rectangular structure, scalability and so on. Attention is focused on designing the routing algorithm CFRA (Cycle free routing algorithm), which is partially adaptive, minimal, deadlock and livelock free. For XD mesh, no virtual channels are required and four virtual channels are needed if wraparound connections are used. Proof of deadlock freedom and livelock freedom is presented. Finally, the simulation is carried out on (8,8)-XD networks to evaluate the performance. The results show that XD network has potentially an attractive interconnection network topology and CFRA is of good performance.
    Chinese Journal of Electronics. 01/2005; 14(4):661-665.
  • Yang Su, Zhanqi Xu, Zengji Liu
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper proposes a dynamic routing optimal model in IP/MPLS over WDM networks. A corresponding on-line integrated routing algorithm minimizing the cost of total lightpaths (MCTLP) is designed accordingly. MCTLP optimizes the network resource utilization by integrally assigning bandwidth allocations on both IP layer and wavelength links usage on WDM layer. The performance comparison of MCTLP with two representative routing algorithms used in IP/MPLS over WDM networks via simulation shows that MCTLP can use less links on IP layer and WDM layer to carry IP traffic flows, resulting in accepting more connection requests and reducing the traffic blocking probability efficiently.
    Sixth International Conference on Parallel and Distributed Computing, Applications and Technologies (PDCAT 2005), 5-8 December 2005, Dalian, China; 01/2005
  • [Show abstract] [Hide abstract]
    ABSTRACT: To make the exiting fault routing algorithms tolerate concave fault regions without disabling any healthy nodes, the concept of hole is proposed in this paper. A hole consists of healthy nodes in the concave parts and neighborhood of a given concave fault region. By guiding the packet routing inside and outside the hole, the new routing method empowers the convex fault tolerant routing algorithm to tolerate concave shape regions without disabling any healthy nodes. The proposed modification method is simple and does not add new virtual channels. Moreover, it doesn't change the rules of the previous algorithms. Finally, the performance of the modified routing algorithm is simulated under various concave fault patterns. 2005 IEEE.
    6th International Conference on Parallel and Distributed Computing, Applications and Technologies, PDCAT 2005, December 5, 2005 - December 8, 2005, Dalian, China; 01/2005
  • [Show abstract] [Hide abstract]
    ABSTRACT: A single rate multicast congestion control for streaming media applications called Binomial congestion control At Receivers for Multicast (BARM) is proposed. Combining aspects of window-based and rate-based congestion control, the protocol shifts most of the congestion control mechanisms to multicast receivers. The main features of BARM are as follows. (1) The protocol adopts binomial algorithm (k=l=0.5, alpha=0.28, beta=0.2 for our implementation) to adjust congestion window, which not only provides TCP-friendliness but decreases abrupt rate fluctuations, making it suitable for real time streaming media multicast applications. (2) The binomial algorithm is executed at the receivers instead of at the sender; to do this, a congestion window is maintained and updated separately by each receiver. Hence the protocol not only has a better scalability but reduces the burden of the sender significantly and is suitable to Client/Server model. (3) The congestion window is converted to the expected receiving rate which is then fed back to the sender if permitted. Compared to window feedback scheme, rate feedback scheme is simpler and increases the scalability. (4) The representative approach is used to suppress the feedback implosion. Simulations results indicate that BARM shows good fairness, TCP-friendliness, smoothness, scalability, and acceptable responsiveness.
    Proc SPIE 10/2004;
  • Source
    Zhen Chen, Zengji Liu, Zhiliang Qiu
    [Show abstract] [Hide abstract]
    ABSTRACT: Cyclic Cubes, proposed as a new family of interconnection networks, is proved not new and is equivalent to the wrapped butterfly networks. There is a similar fact that the Cyclic Cubes is also equivalent to the Bidi Shufflenet, which is investigated recently. But this fact is almost neglected. In this paper, we discuss the equivalence between the Cyclic Cubes and the Bidi Shufflenet.
    Advanced Information Networking and Applications, 2003. AINA 2003. 17th International Conference on; 04/2003
  • Hao Qin, Shi Zhang, Zengji Liu
    [Show abstract] [Hide abstract]
    ABSTRACT: We propose a dynamic routing and wavelength assignment algorithm in an all-optical network with limited-range wavelength conversions. For each randomly arrived request, the correlation of different wavelengths on different links in all possible routes is described by the wavelength weight in our algorithm. We try to assign wavelengths on each candidate route using a shortest-path algorithm based on the wavelength weight and choose the best as its route and wavelength assignment scheme. Numerical results obtained for NSFNET and a 12-node ring show that our algorithm can greatly reduce the blocking probability and the number of required conversions under uniform traffic or traffic dominated by larger hop routes.
    IEEE Communications Letters 04/2003; · 1.16 Impact Factor
  • Hao Qin, Zengji Liu, Shi Zhang, Aijun Wen
    [Show abstract] [Hide abstract]
    ABSTRACT: In this letter, the routing and wavelength assignment problem in all optical networks is considered. We improve the Max-RWA model presented by R. M. Krishnaswamy and K. N. Sivarajan (see ibid., vol.5, p.435-7, Oct. 2001) by introducing limited-range wavelength conversions. Considering transmission performance, we propose a new optimization objective, which is to establish the maximum number of connections with the least number of wavelength converters. The modified Max-RWA problem is formulated as a integer linear programming (ILP) problem, and then solved using a genetic algorithm. The extended layered-graph approach is used to assign routes and wavelengths when necessary, thus reducing the complexity of the genetic algorithm. Numerical results obtained for NSFNET are presented.
    IEEE Communications Letters 11/2002; · 1.16 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Aggregating the traffic to different forwarding equivalence class is an indispensable and significant phase for layer-3 and layer-2 hybrid switching. Regarding existed schemes mainly focusing on high efficiency of network resource but taking little consideration on user's demand, we put forward a new scalable framework for traffic partitioning based on some generalized concepts and optional priority serving mechanism to satisfy both users and networks' requirement. In this letter, the design of the conceptual framework regarding the key technologies and corresponding adaptive algorithms are presented. Simulation using real Internet traces shows that our adaptive adjustment can guarantee delivering bound, the maximum delay of concerned packets while consuming bandwidth around the average traffic rate.
    IEEE Communications Letters 05/2002; · 1.16 Impact Factor
  • Hua You, Zhiliang Qiu, Zengji Liu
    [Show abstract] [Hide abstract]
    ABSTRACT: Implementing QoS, one of the eternal research subjects, on Internet has been attracting a lot of attentions although several schemes are already available. This letter proposes a new service architecture—Scalable Tri-tier Service Architecture (STSA) and a fast label switching scheme, which can reduce the overhead of signaling and suit the applications of access demand for a great number of end users with QoS guarantee. Also presented in the letter is an implementation of STSA in MPLS scenario.
    Journal of Electronics (China) 01/2002; 19(4):415-419.