Computer Networks Laboratory

/ Zi Xue Cheng / Associate Professor
/ Vyacheslav Tuzlukov / Visiting Professor
/ Akio Koyama / Assistant Professor

Computer Networks Laboratory is working in six directions within Research Projects.

1. Logical-timing and Decentralized Control in Computing Systems;
2. Beta-CMOS implementation of artificial neuron;
3. Network-based Computing and Distributed Computing;
4. Network Agents;
5. Network Applications.
6. A Media Access Protocol for Super High Speed Networks.

1. Logical-timing and Decentralized Control in Computing Systems

One of the main problem of this direction is Global Synchronization of Asynchronous Arrays in Logical Time. It is shown that the duration of current transient process in CMOS circuits depends on the number of concurrently switching gates k, increasing approximately by $ln k$. It makes doubtful the efficiency of using current sensors in circuits with high concurrency. A class of circuits is discussed for which efficient usage of current sensors is suggested.

2. Beta-CMOS implementation of artificial neuron

2.1 The improved version of digital-analog CMOS implementation of an artificial neuron is suggested. It is built on the basis of the circuit consisting of synapses, $\beta$-comparator and output amplifier. It has been shown that higher non-linearity of the $\beta$-comparator in the threshold zone can sharply increase the threshold of the realized functions and noise-stability of the neuron.

2.2 The functional possibilities, parameter stability, and learnability of the artificial $\beta$-CMOS neuron have been considered. The SPICE simulation results confirm that the neuron is learnable to realize threshold functions up to 12 variables.

2.3 One of the problems in designing analog/digital artificial neuron is answering the question about what class of threshold functions can be used for testing the learnable neuron. It was shown that Gorner's threshold functions cover wide range of input weights, have big value of the threshold, and can be recommended for usage as test functions.

2.4 The earlier suggested neuron implementation contains only synapses with excitatory inputs. However, most problems solved by artificial neural networks either require inhibitory inputs. Thus the neuron should have synapses capable to form the weight and type (excitatory or inhibitory) of the input during the learning. Several synapse circuits are suggested with two capacitors for storing positive and negative input weights.

3. Network-based Distributed Computing

3.1 Distributed Resource Allocation among Overlapping Groups
In this research, we presented a new problem which described explicitly the competition for resources among process groups which may share common processes. A solution which allocates resources to groups of processes with deadlock among groups and starvation of a group never happening, was given.

3.2 Distributed Algorithms for Leader Election on Partially Ordered Keys
In this research, we generalize the traditional leader election problem, in the way that finds all the processors with the maximal keys on the basis of a partial order on the keys, and proposed two distributed algorithms for the generalized problem.

4. Network Agents

4.1 Match-making network agents which consider psychological factors of users
Match-making is a service which helps users to find suitalbe friends or partners in network environment. We developed methods for match-making, with some psychological factors of users being considered.

4.2 Learning assistant agents
In this research, we were developing teaching assiatant agents which can catch not only the understanding states of learners but also psychological states of the learners, in order to provide more efficient learning support to learners.

5. Network Applications

The goal of the project is to develop an interactive and personalized tele-education environment over Gigabit network by applying the state-of-the-art technologies in high-speed network, multimedia delivery, and intelligent software agents. Some core parts of the system were implemented and some real time online tele-lecture experiments were performed.

6. A Media Access Protocol for Super High Speed Networks

In this work, we develop a media access protocol for terabit networks such as Wavelength Dvision Multiplexing (WDM) networks. The WDM networks are promising networks which will be used in future Internet. Our goal is to design a protocol with high throughput, low delay, fairness and be used for multimedia communications.

Refereed Journal Papers

1. A. Koyama, L. Barolli and Z. Cheng., An Adaptive Media Access Protocol for Ring Topology Terabit Networks. IPSJ Journal, 2001, Vol. 42, No. 6, pp. 1676--1683.

2. L. Barolli, A. Koyama, T. Yamada and S. Yokoyama., An Integrated Fuzzy Policing-Routing Mechanism for ATM Networks Using Violation Tagging Function. IPSJ Journal, 2000. vol. 41, No. 10, pp. 2926--2936.

3. L. Barolli, A. Koyama, T. Yamada, and S. Yokoyama. An Intelligent Policing-Routing Mechanism Based on Fuzzy Logic and Genetic Algorithms and Its Performance Evaluation. IPSJ Journal, 2000. vol. 41, No. 11, pp. 3046--3059.

4. L. Barolli, A. Koyama, T. Yamada, and S. Yokoyama. An Integrated CAC and Routing Strategy for High-Speed Large-Scale Networks Using Cooperative Agents. IPSJ Journal, 2001. vol. 42, No. 2, pp. 222--233.

5. A. Koyama, L. Barolli, S. Mirza and S. Yokoyama. Performance Evaluation of Self Detective Congestion Control Scheme for ATM Networks. Advanced Information Processing Technology Series, Information Networking in Asia, G&B publisher, 2001. vol. 3, No. 2, pp. 25--36.

6. A. Koyama, L. Barolli and Z. Cheng., An Adaptive Media Access Protocol for Ring Topology Terabit Networks. IPSJ Journal, 2001, vol. 42, No. 6, pp. 1676--1683.

1. Zixue Cheng, Yutaka Wada, Yao Xue Zhang, and Shoichi Noguchi. Distributed Solution for Resources Allocation to Overlapping Groups. Seventh International Conference on Parallel and Distributed Systems (ICPADS'2000), Jul. 2000.

2. A. He, Z. Cheng, T. Huang, R. Nakatani, Y. Amadatsu, A. Koyama, Y. Zhao, and S. Noguchi. Design of Real-Time Interactive Tele-exercise Classroom for Computer Exercises over a Gigabit Network. ICOIN'01, 2001, pp. 757-763.

3. A. Koyama, L. Barolli, A. Tsuda, and Z. Cheng. An Agent-Based Personalized Distance Learning System. ICOIN'01, 2001. pp. 895-899.

4. A. Koyama, L. Barolli, and Z. Cheng. Performance Evaluation of an Adaptive Ring Access Protocol for Super High-Speed Networks. ICOIN'01, 2001, pp. 201-206.

5. Z. Cheng, Y. Wada, S. Hashimoto, A. He, T. Huang. A New Method for Constructing Efficient Local Coteries. ICOIN'01, 2001, pp. 512-517.

6. Z. Cheng, A. He, T. Huang, A. Koyama, S. Noguchi, N. Honda, Y. Shibata, and N. Shiratori. An Overview of an Interactive and Personalized Multimedia Tele-Education Environment over a Gigabit Network. IS2000, 2000. pp. 567-571.

7. Aiguo He, Zixue Cheng, Tongjun Huang, Ryohei Nakatani, Yuichi Amadatsu, Akio Koyama, Yue Zhao, and Shoichi Noguchi. Design of a Tele-Exercises Classroom for Computer Exercises over the Gigabit Network. IS2000, 2000. pp. 572-575.

8. Guozhen Zhang, Ken'ichi Saitou, Akio Koyama, and Zixue Cheng. Agent-Based Methods for Course Contents Construction and Effective Learning Support. IS2000, 2000. pp. 496-500.

9. Aiguo He, Zixue Cheng, Tongjun Huang, Ryohei Nakatani, Yuichi Amadatsu, Akio Koyama, Yue Zhao, and Shoichi Noguchi. Design of Distance Education System for Computer exercises over a Gigabit Network. The 8th workshop on Multimedia Communication and Distributed Processing, 2000, pp. 13-18.

10. Guozhen Zhang, Ken'ichi Saitou, Akio Koyama, and Zixue Cheng. A Proposal of Agent-Based Construction and Presentation Methods of Active Courseware According to the Situations of Learners. The 8th workshop on Multimedia Communication and Distributed Processing, 2000. pp. 67-72.

11. Ken'ichi Saitou, Guozhen Zhang, Akio Koyama, and Zixue Cheng. A Method for Construction of Personal Learning Courses Using Agent. The 8th workshop on Multimedia Communication and Distributed Processing, 2000. pp. 73-78.

12. Shoichi Nakamura, Kanako Saitou, and Zixue Cheng. Proposal of Education Support Environment for Personalized Education. The 8th workshop on Multimedia Communication and Distributed Processing, 2000. pp. 19-24.

13. Satoru Hashimoto, Yutaka, Wada, and Zixue Cheng. local Majority Coteries based Distributed Algorithm for Resource Allocation. The 8th workshop on Multimedia Communication and Distributed Processing, 2000. pp. 247-252.

14. L. Barolli, A. Koyama, T. Yamada, and S. Yokoyama. A Comparative Analysis of Two Genetic Algorithm Based Routing Methods for High-speed Networks. 2nd Asia-Pacific Conference on Genetic Algorithms and Applications, pp. 332--341, Global-Link Publishing Company, May 2000.

15. L. Barolli, A. Koyama, T. Yamada, and S. Yokoyama. An Intelligent Call Admission Control Scheme Based on Fuzzy Logic. World Multiconference on Systemics, Cybernetics and Informatics, pp. 84--89, July 2000.

16. L. Barolli, A. Koyama, T. Yamada, and S. Yokoyama. An Intelligent Fuzzy Routing Scheme for Improving ATM Networks Performance Using Violation Tagging Function. The eleventh International Workshop on Database and Expert Systems Applications / Network Based Information System, pp. 5--9, IEEE Computer Society, Sep. 2000.

17. A. Koyama, L. Barolli, S. Yokoyama, and Z. Cheng. An Agent-based WWW Mirror Server Selection Method. IEEE International Workshop on Intelligent Signal Processing and Communication Systems, pp. 969--972, IEEE, Nov. 2000.

18. L. Barolli, A. Koyama, G. Capi, T. Yamada and S. Yokoyama. An Intelligent Routing Framework for High-Speed Large-Scale Networks Using Cooperative Agents. IEEE International Workshop on Intelligent Signal Processing and Communication Systems, pp. 963--968, IEEE, Nov. 2000.

19. A. Koyama, L. Barolli and Z. Cheng. Performance Evaluation of an Adaptive Ring Access Protocol for Super High-Speed Networks. 15th International Conference on Information Networking, pp. 201--206, IEEE, IEEE Computer Society, Jan. 2001.

20. A. Koyama, L. Barolli, A. Tsuda and Z. Cheng. An Agent-based Personalized Distance Learning System. 15th International Conference on Information Networking, pp. 895--89, IEEE, IEEE Computer Society, Jan. 2001.