Jung Wan Cho

A design of macro-micro CDMA cellular overlays in the existing big urban areas

We focus on delivering radio engineers a practical design of macro-micro code division multiple access (CDMA) cellular overlays. First, we review our algorithmic approach to jointly deploy macrocells and microcells over todays big urban areas having spatially nonuniform traffic distributions. Next, we identify several further issues related to the optimal design of macro-micro cellular overlays and enhance the cell-deploying algorithm to reflect these issues. The numerical results by extensive event-driven simulations show that the resulting macro-micro cellular overlays successfully cope with the existing conditions of todays big urban areas, such as spatial and temporal traffic distributions and user mobility characteristics. Finally, we discuss the practical guidelines for designing macro-micro cellular overlays in the existing big urban areas.

A fast handoff scheme for IP over Bluetooth

A lot of people use a cell phone, PDA, and laptop computer in their daily lives. Bluetooth which enables incompatible devices to communicate with each other over a wireless link can speed up the growth in popularity. BLUEPAC(BLUEtooth Public ACcess) is designed to support IP over Bluetooth and provide users with access to a public network such as the Internet. Our work is motivated by the need for a fast handoff in BLUEPAC A mobile device experiences frequent handoffs due to the short radius of the piconet which is about 10 m. A handoff requires a connection establishment between a mobile device and a new base station which is is a time-consuming procedure in the frequency hopping system. This situation may cause many connection losses which hurt the network performance as a result. We present a new scheme for improving the handoff performance. The basic idea is that a base station controller helps a mobile device to establish a connection to new base station by giving their addresses and clocks. Simulation results show that our scheme significantly reduces handoff delay and imposes a very light overhead on the base station.

Application of fully recurrent neural networks for speech recognition

The authors describe an extended backpropagation algorithm for fully connected recurrent neural networks applied to speech recognition. The extended delta rule is approximated by excluding some of the past activities of the dynamic neurons to reduce computational complexity without performance degradation. In speaker-dependent recognition of a confusable syllable set, the fully recurrent neural network with the approximated backpropagation algorithm showed better performance than the multilayer perceptron and the self-recurrent network with comparable time complexity. In addition, it is found that most self-recurrent connections become excitatory and most mutual recurrent connections become inhibitory.<<ETX>>

Secure group communication with multiplicative one-way functions

Secure group communication enables only the users in that group securely communicate with each other. Because the users can join or leave the group dynamically, scalability is a major concern. In this paper, we propose a communication-efficient model using multiplicative one-way functions. It also has well known tree-based structure but uses exclusive keys that are held by the users not in some subtree. When totally n users are in the group, proposed model reduces the size of multicast message required in join or leave operation from O(logn) to O(1) keeping other costs comparable. The security of proposed model is based on the difficulty of finding square root modulo composite number and it also satisfies both forward and backward secrecy. Therefore, proposed model is very applicable to a wide area network environment or a low-bandwidth channel such as ad hoc network.

A heuristic processor allocation strategy in hypercube systems

A new processor allocation scheme for hypercube systems, called the HPA (heuristic processor allocation) strategy, is presented. In this scheme, an undirected graph, called the SC-graph (Subcube-graph), is used to maintain the free subcubes available in system, which are represented by vertices. An allocation request for a k-cube is satisfied by finding a free subcube of dimension k in the SC-graph or by decomposing a nearest higher dimension subcube. If there are more than one subcube of dimension k, a subcube which has minimum degree in the SC-graph is selected to reduce the external fragmentation. For deallocating the released subcube a heuristic algorithm is used to maintain the dimension of free subcube as high as possible. It is theoretically shown that the HPA strategy is not only statically optimal but also it has a complete subcube recognition capability in a dynamic environment. Extensive simulation results show that the HPA strategy improves the performance and significantly reduces the allocation/deallocation time compared to the previously proposed schemes.<<ETX>>

Design of rate-based congestion control scheme for MPEG video transmission in ATM networks

In this paper, we investigate MPEG video transmission over available bit rate (ABR) service. It requires small cell delay variation and non-trivial minimum cell rate (MCR) to guarantee quality of service (QoS). Existing ABR congestion control schemes cannot meet the requirements since they are designed for data applications. We propose a closed-loop rate-based congestion control scheme called explicit rate indication scheme for MPEG (ERISM). ERISM maintains the queue length minimal, and allocates cell rates to connections with non-trivial MCR. Simulation results show that ERISM provides a good transmission capability for MPEG video.

A logic programming approach to hybrid knowledge representation

A new hybrid system called Sphinx is presented which combines classification-based reasoning with logic programming technique. It consists of two major components: a terminological component and an assertional one. The former uses a frame-based knowledge representation scheme based on classification and the latter is a theorem prover based on Horn logic. The assertional component can also handle a first-order query language and wh-questions. A new knowledge base maintenance mechanism based on the negation as failure inference rule enables Sphinx to support incremental assertions and retractions. In addition, Sphinx provides explanation capability to help the user in developing and debugging knowledge bases.

A graph-based subcube allocation and task migration in hypercube systems

The authors propose a task migration scheme based on the HSA (heuristic subcube allocation) strategy to solve the fragmentation problem in a hypercube. This scheme, called CSC (complementary subcube coalescence), uses a heuristic and an undirected graph, called the SC (subcube) graph. If an incoming request is not satisfied due to the system fragmentation, the task migration scheme is performed to generate higher dimension subcubes. Simulation results show that the HSA strategy gives better efficiency than the Buddy and GC strategies in the adaptive mode. Moreover, the HSA strategy has a significantly lower migration cost than that of the Buddy and GC strategies.<<ETX>>

A Cluster-Based Router Architecture for Massive and Various Computations in Active Networks.

Traditional network routers are passive in a manner of speaking because they can not manipulate packets but just deliver them. On the contrary, an active network is an innovative approach to the network architecture. Active routers can perform customized computations on the packet flowing through them. They thus can easily adopt or remove protocols and perform operations such as firewall, content-based switching, and multimedia broadcasting. However, they may suffer from the lack of computational power in the near future due to growing active applications. Cluster-based active router can be a solution of this problem. A cluster system is easier to build than a multiprocessor system and brings out similar computing performance. We designed a cluster-based router architecture to share the overall computational load. In our design, packets can be forwarded to other nodes within the cluster using high-speed interconnect network. Therefore, busy node’s load can be transferred to other nodes quickly. To test the feasibility of our cluster-based active router design, we modified existing active router daemon, Anetd, so that it can work on the cluster and forward packets inside the cluster. After performing some micro-benchmark, we observed that overall latency is reduced.

Design and implementation of a multi-threaded TMN agent system

A telecommunications management network adopts the OSI manager-agent model in which an agent system plays a more important role than a manager system. Because the existing agent systems sequentially perform a CMIS operation on selected managed objects, it cannot guarantee quick response time when many managed objects are selected. We propose a multi-threaded process structure for the agent system which performs CMIS operations concurrently on managed objects. We implemented our agent system using OSIMIS-4.O TMN platform in Solaris. We measured response time to a CMIS operation to evaluate the performance of our system. Experimental results show that the performance of our system is not so good when network resources reside within the agent system, but very good when network resources reside outside the agent system.