In this paper, a spectral and power efficient multicarrier modulation technique named MC-SQAM (Multi-Carrier Superposed -QAM) is introduced for use in a nolinearly amplified transmission channel. The probability of error performance of MC-SQAM in a non-linearly amplified channel is experimentally anlayzed via computer simulation in the presence of additive white Gaussian noise (AWGN). It is found that the CNR improvement of the proposed MC-SQAM compared to a conventional MC-QAM operating at 12bB IBO of Hughes TWTA is 3.5dB at Pe=2x10-4. It is also found that thd CNR improvement of MC-SQAM compared to a conventional MC-QAM operating at 3dB IBO of Fujitsu SSPA is 4.3dB. The result shows that MC-WQAM would achieve higher efficiencies of spectrum and power utilization than conventional MC-QAM in a severely bandwidth and power limited wireless multimedia system.

In this paper, we studied the performance of a proposed decoder with a general optimum decoder in a severe intersymbol interference(ISI) channel. A linear transversal equalizer is used to overcome ISI. With three different block codes, proposed decoder using a binary neural network showed a similar performance to the optimum decoder. The proposed binary neural network has several merits. It was only integer values in their coefficients and thresholds and uses a hard-limiter as a neuron. So it is very easy to implement the high-speed decoder. Also, it is possible to train the binary neural network as various types of decoders. A binary neural network, therefore, can be used as a general purpose decoder in a single chip. The binary neural networks are trained with the geometrical expansion learning algorithm which guarantees convergence for any binary-to-binary mapping. In this paper, training approach, i.e. the off-line training approach is simulated and its results are included.

In spite of growing importance of input queueing ATM switches, there have been little research on performance of input queues, of ATM switches, to which combined priority controls of space and time priority schemes are applied. In this paper, assuming two classes of traffics requiring high and low quality-of-service(QoS), we analyze a combined priority control method of partial buffer sharing and general state-dependent scheduling schemes in the put queues of a nonblocking asynchronous transfer mode (ATM) switch. The analysis is done in two steps, using a semi-Markov process concept. As results, we show how various system parameters affect the performances. Particularly, we show how the threshold value and the scheduling parameter of the combined scheme can affect dynamically the performance differences between the two classes. We further show how we can adjust the threshold and the scheduling parameter against a wide range of QoS gaps between the two classes to maximize the admissible load.

In this paper, we propose medium access control(MAC) protocol for international mobile telecommunication system-2000(IMT-2000) that can support both voice and data service. In the proposed protocol, both voice and data traffic transmitted in reservation mode, which is originally proposed for voice traffic transmission. Transmitting data traffic in reservation mode can reduce the frequency of reservation procedure of each data terminal. Simulation is performed to find system parameters and to evaluate system performance. Simulation result shows that proposed protocol is appropriate for integrated voice/data system and is superior to other systems which are not using reservation scheme for data traffic transmission.

We consider frequency overlapped DS/CDMA (FO/CDMA) communication systems where a number of DS/CDMA systems share frequency bands with adjacent systems. We analyze the multiple-access interference from adjacent systems with arbitrary amount of frequency overlap and compare the capacity of FP/CDMA with that of single wideband CDMA(WCDMA). The optimum amount of overlap increases with the number of overlapped systems. However the maximum capacity is obtained when the optimum number of systems are overlapped by half null-to-null bandwidth. It is shown that the FO/CDMA yields higher capacity in the case of relaxed BER requirement and low decay rate of multipath intensity profile. Otherwise, WCDMA is superior.

In this paper, the BER, CLR, and CER of wireless ATM cell adopting error control coding and diversity reception techniques have been derived theoretically in a fading channel for a design of effective wireless ATM system. The amount of improvement in throughput of wireless ATM system dopting the FEC/truncated Type-Ⅱ hybrid ARQ technique has been evaluated. The obtained results confirm that the adoption of MRC diversity technique is more effective than using SC diversity technique. It is also confirm that convolutional coding technique gives better performance than the other coding techniques. When we adopt the truncated Type-Ⅱ hybrid ARQ technique as DLC layer protocol of wireless ATM, the throughput has improved about 15% as compared with Type-Ⅰ hybrid SR-ARQ technique.

In this paper, a dynamic band width allocation based in transfer rate ratio is proposed to efficiently provide ABR service in ATM networks. The proposed method is based on the bandwidth allocation of a distributed Explicit Rate(ER) in ATM network. The ABR source estimates a Segmental Average Source Cell Rate(SASCR) for every block size of Nrm cells and sends it to ATM networks using forward RM cell. When ATM switch receives the forward RM cell, it computes the sum of the cell rate that each ABR source uses. On receiving the backward RM cell, ATM switch computes the bandwidth to be allocated to each ABR source based on cell transfer rate ratio within the limits of the available bandwidth of ATM networks. The propose method for bandwidth computation in ATM switch guarantees MCR which ABR source negotiates with ATM networks. Since the proposed method performs the bandwidth allocation based in cell transfer rate ratio of each ABR source, ATM switch can rapidly adapt to the bandwidth variation of ATM networks and the transfer rate variation of ABR source. Thus the method makes the best use of available bandwidth in ATM networks. The simulation results demonstrate that the proposed that the proposed method achieve the more fair bandwidth allocation and higher link utilization than those achieved by existing methods.

This paper presents the design of the RBAC (Role-Based Access Control) for security systems. Security system interest have exploded in the last few years. But, on existing security frameworks are lack and many problems. Therefore, we propose to design of a RBAC for a new security systems. RBAC is a new technology that is attracting increasing attention, particularly for commercial applications, because of its potential for reducing the complexity and cost of security administration in large network based applications. The RBAC components can be linked with commercially available application servers, and require no modification of the server software.

The new network architecture framework for the novel services such as the virtual network services, the mobile and multimedia service requires more flexible access and management than current network infrastructure are capable of providing. To meet the customers' or users' requirement for these services, it requires a network infrastructure into which services can be introduced easily, openly, and quickly. Interoperability, portability and reuse of software components will have prime importance in this new software architecture framework. The software based infrastructure that is called Telecommunication Information Networking Architecture (TINA), would be a main stream for enabling services. In this paper, for supporting customized mobile services, we adopt a new service component, which we call UpeMgr (Ubiquitous Personal Environment Manager). In order to support mobile users who move between heterogeneous networks, for instancess between the TINA-like environment and the Internet environment, we propose a structure of gateway. The proposed UpeMgr uses the fixed and mobile agent approaches for supporting the user's mobility, and we evaluated their performances by comparing those approaches. The proposed UpeMgr object supports personal mobility and session mobility for the mobile users in telecommunication information networks.

This paper discusses problems arising out of the use of Mobile Internet Protocol (IP) in a security conscious Internet with packet filters and firewalls. An enhanced firewall traversal solution for Mobile IP is proposed in case there are firewalls protecting visited networks as well as home networks of mobile nodes. While previous solutions are asymmetric ones in the sense that only home networks are protected by firewalls, this solution is a symmetrical one. Also, packets from a mobile node can go directly to the correspondent node via a "Open Node" just outside of a foreign domain without going through the home agent of the mobile node. Therefore the solution uses an almost optimal routing. This is achieved with minimal overhead such as using open nodes and adding necessary headers to packets directed from the mobile node to the correspondent node.

In this paper, we propose a new consolidation algorithm at branch-point switch for point-to-multipoint ABR (Available Bit Rate) flow control in ATM networks and compare is performance with existing consolidation algorithms. We first investigate the consolidation problems of existing algorithms, and then propose some solutions for resolving those problems. By combining these solutions, we propose a counter-based consolidation algorithm. The proposed algorithm stores the feedback information on a per-branch basis, and maintains a counter in order to distinguish the BRM (Backward Resource Management) cells returned from the farthest destination. Furthermore, we compare the performance of the proposed algorithm with the existing algorithms under various network environments. Them simulation results show that the proposed algorithm exhibits a better performance than the existing algorithms in terms of allowed cell rate oscillation, queue length, utilization, and fairness.

In this paper, one forecasting method for short-term forecasting, a market survey is introduced which will be used for the estimation of subscriber line demand in the optical access networks. The market survey method for the new multimedia services is attempted to collect information directly from customers using the questionnaires for home-users and business-users in local loops. Analysis rationale of questionnaires is suggested to estimate the number of subscriber lines. Also, two measures are presented to estimate it one is the probability that the customer will use the multimedia services and the other is the rate that the subscriber line demand will be actually realized. The former measure is calculated based on the customer information using the logit analysis. The latter is obtained by the degree of customer's knowledge about the specific and the customer's willingness to use the services.

RMON probe attached in a network operates in promiscuous mode and statistics information in each segment and can provide proactive monitoring, host statistics, matrix statistics, user-defined alarm, event generation and packet filtering/capture. The control of RMON probe and the analysis of management information are required to extract performance parameters from RMON MIB. In this paper, we have defined the LAN performance parameters for LAN management using RMON MIB. Also, we have proposed the calculation model and algorithm of the performance parameters. Each algorithm includes the numerical formula for the parameter and RMON probe control method for collecting management information on LAN. To study the characteristics of the suggested algorithm, we have tested them in a real LAN and proved the adaptation and the reliability. LAN performance parameters, which have proposed in this paper, will enable a network manager to estimate and diagnose the status of a LAN and use the available information as and indicator for LAN design and configuration.

A simple routing algorithm such as self-routing plays a critical role in an ultra-fast multihop packet switching network. To achieve this property, a variety of regular network topologies has been proposed. Among them, the Shufflenet has been modified and generalized by several researchers to alleviate the constraint on the number of realizable nodes. In this paper, a new network architecture based on the Sufflenet is presented. This network relieves the constraint on the number or nodes realizable greatly, but still exploits the simple self-routing algorithm. Deflection routing can be combined with self-routing in the proposed network, and even when a packet is deflected, the header information can still be reused without modification.

In this paper, an adaptive array algorithm is presented to minimize linear crosstalk in fast WDM system and the crosstalk levels are monitored by using the pilot tone which is unique to each channel. The crosstalk levels of other adjacent channels can be determined by the power levels of pilot signal and the weighting and summing the photocurrent of channels can cancel the linear crosstalk. Also, it can compensate the drift of laser frequency using pilot tone as reference signal. The speed of algorithm processing reaches that of transmittance in fast WDM by sampling a bit per finite bits.

Mode convertible, polarization-independent optical filters at 1.55㎛ wavelength have been fabricated by polarization mode converters utilizing Ti:LiNbO₃ optical waveguides and strain-optic effect, and electrooptic effect to change the birefringence. The polarization mode converters was realized by utilizing the shear strain from an evaporated 1.5㎛ periodic SiO₂ thin film at 300℃, and electro optically tunning the wavelength for maximum polarization conversion via the electrooptic effect. The 12dB TM→TE mode conversion was measured at λ=1564nm for the having 9㎛ waveguide width and 400 periodic pads with 19.4㎛ period.

In this paper, an embedded wavelet coder, called a band-selective embedded wavelet coder, is proposed. The well-known embedded zerotree wavelet (EZW) coder uses the successive approximation quantization (SAQ) process and zerotree structures of wavelet coefficients. The EZW coder scans iteratively all the wavelet coefficients during the SAQ process, which decreases the coding efficiency considerably. In the proposed scheme, we scan only significant subbands using the significance test with the multi-threshold. The multi-threshold obtained from the coefficient with the maximum magnitude in each subband is used to determine whether a subband has significant coefficients or not. Since it is not necessary to encode the insignificant subbands having no significant coefficients during the SAQ process, the redundancy generated by scanning higher subbands can be significantly reduced. Experimental results show that the proposed method outperforms popular image coders such as JPEG and EZW.

In this paper, we seek to find an optimal allocation of the link capacity of Network Access Subsystem(NAS) of Advanced Information Communication Processing System(AICPS). An effective partition of limited link capacity decrease NASs call blocking probability. We analyze users call behavior by the first step analysis, and develop a steady state balance equation of the user behavior. Then, we use the overrelaxation method to solve the problem. Implementing the proposed solution to the data network system, we expect a significant enhancement of AICPS performance.

In this paper, a new method for effective video coding is studied. Original video frames are transformed into hierarchical pyramidal images with multiresolution between neighboring frames is obtained from the low-resolution band. Generally, most of the energy distribution of this pyramidal structure resides in the low-resolution band. Also, to control the video coding procedure, a new picture set filter is proposed. This filter controls the compression ratio of each frame depending on the correlation to the reference frame by selectively eliminating less important high-resolution areas. Consequently, video quality can be preserved and bit rate can be controlled adaptively. In the simulation, to test the performance of the proposed coding method, comparisons with the full search block matching algorithm and the differential image coding algorithm are made. In the former case, video quality is improved by 9dB, compression ratio is increased by 15 times and coding time is reduced by 15%. In the latter case, video quality is degraded by 25dB, but compression ratio is increased by 42 times and coding time is reduced by 0.09 %. Consequently, the proposed method shows a reasonably good performance over existing ones.

In this paper, an effective block-based image compression technique using wavelet transform for high compression rate and low error rate is proposed. The proposed algorithm decomposes the input image into several subimages using the 2-D discrete wavelet transform and rearranges subimages into small blocks for compression. The algorithm uses the human visual characteristics to obtain better performance. Experimental results show that the proposed algorithm achieves better performance in terms of RMSE, PSNR, and BPP and also show better subjective quality than the existing block-based image compression techniques which do not use wavelet transform. In addition, we obtained subjective improvement in the reconstructed image by reducing blocking effects.

This paper presents a new indexing technique, for image retrieval, which calculates the similarity of distance between projected DC histogram result from the same process. By DCT, which is essential procedure in block-based JPEG and/or MPEG standard, DC image can be constructed directly in compressed domain and this represents the average of block segmented image. We can obtain the positional information by projection with horizontal and vertical direction, and map this information into histogram that is used as database indices. Distance approximation of the projected DC histogram is used as a matching criterion. From the simulation results, the proposed method clearly show the validity and superiority in respect of computation time and memory space and that in conjunction with other techniques for indexing, such as color, can provide a powerful framework for image indexing and retrieval. Furthermore, the proposed method can be applied to video scene change detection usefully.

The variable structure control(VSC) with sliding mode is an important and interesting topic in modern control of nonlinear systems. However, the discontinuous control law in VSC leads to undesirable chattering in practice. As a method solving this problem, in this paper, we propose a scheme of the VSC with neural network sliding surface. A neural network sliding surface with boundary layer is employed to solve discontinuous control law. The proposed control scheme is verified by simulation results.

In this paper, the predictive pyramid VQ(PPVQ) is proposed to quantize LSF parameters. The properties of the LSF parameters, whose prediction error signal is Laplacian and interframe correlation is high, are exploited. To predict the LSF parameters, 4th order moving average (MA) predictor is used, and to quantize the prediction error signal, the pyramid VQ (PVQ) is exploited. It is verified that the performance of PPVQ has the transparent quality at 23 bits/frame. PPVQ requires low complexity and does not need any memories for the codebook storage.

This paper concerns with improving the previously developed RBF equalizer by greatly reducing the number of centers. The basic idea is to select only centers close to the boundary between the different decision classes. The first factor of reducing the network is where dis the channel delay. The number of centers was further reduced by representing several centers by a single point. This reduction of centers greatly reduces the burden of computation in training, and makes the hardware implementation of RBF equalizers realistic. Simulation studies show that the error rate performance of an RBF equalizer with the propose reduction in the number of centers compares favorably with the RBF having the conventional number of centers.