By Nathan Olson,2014-12-25 23:31
12 views 0

Department of Communication Engineering

    Title : CPW folded quarter-wave resonator filters with transmission zeros Principal Investigator: Chi-Yang Chang

    Sponsor: National Science Council

    Keywords:Folded CPW resonator, CPW filter with finite frequency transmission

    zeros, CPW bandpass filter.

     In this program, a folded CPW (coplanar waveguide) resonator is proposed. This type of resonator can be used to construct various filters such as combline structure, interdigital structure, parallel coupled structure, etc. The folded resonator is shorter than conventional resonator and the size of whole filter can also be shrunken. In the mean time the layout of the filter will also be studied so that the filter may have finite frequency transmission zeros. Preliminary study shows that the cascade quadruplet (CQ) and cascade trisection (CT) structure is very possible to implement finite frequency transmission zeros. The distance between ground plane and internal resonator of CPW combline or interdigital filter is usually too large and often causes layout problems. We propose a method that inserts ground planes between every two resonators. This newly inserted ground plane changes the coupling and needs to study further.



Title : Optical Access Networking Technology(2/3)

    Principal Investigator: Chung-Ju Chang

    Sponsor: National Science Council

    Keywords: Fiber-In-The-Loop (FITL), Optical Burst Switching (OBS), Optical

    Packet Switching (OPS), Active Optical Network (AON), Passive

    Optical Network (PON), Ethernet Passive Optical Network (EPON),

    Traffic Grooming

    For the excellent performance of silica fiber, such as wide bandwidth, low transmission loss, lightweight, and immunity to interference, it is broadly deployed in long haul trunk and is gradually migrated into the access network to form the Fiber-In-The-Loop (FITL) architecture. Many efforts have been made to overcome the economic barrier for the mass deployment of FITL systems since it can improve

    the quality of the network and enhance the flexibility to provide the new broadband services. The rapid spread of the World Wide Web and increasing applications of digital contents have significantly increased have significantly increased the demand for high-speed Internet access. The networking of company LANs for intranets and extranets has also been increasing. They all need the broadband fiber access system and optical network to provide such high bandwidth and services.

    Classical approaches to QoS provisioning in IP networks are difficult apply in all-optical networks. This is mainly because there is no optical counterpart to the store-and-forward model that mandates the use of buffers for queueing packets during contention in electronic packet switches. This project will propose some QoS provisioning mechanisms in IP over metro-access WDM networks. Optical Burst Switching (OBS) network combine the advantage of both wavelength-routing (WR) and Optical Packet Switching (OPS) networks. As in WR networks, there is no need for buffering and electronic processing for data at intermediate nodes. At the same time, OBS increases the network utilization by reserving the channel for a limited period. A signaling protocol or reservation protocol to control shall be used to provide QoS, and a scheduling mechanism is used to resolve the output wavelength contention in OBS nodes.

    As the SDH-based ring networks are evolving into metro-access WDM-Ring networks. The bandwidth request of a lightpath. The project will propose an efficient grooming algorithm for low-speed connections onto high-capacity lightpaths to improve the network throughput and reduce the SDH/WDM network cost.

    The concept of optical Ethernet has popped up toward the access network. The Ethernet Passive Optical Network (EPON) will become a dominant technology for those services offered by EPON may range from simple telephony to multimedia communication. Thus it needs to make a differentiated classification for the specific customers or services. The IEEE 802.3ah Ethernet in the First Mile (EFM) introduces Ethernet LAN, which is a peer-to-peer architecture, applying to a shared media access based PON architecture. That clicks issues for processing and new design of MAC and MAC control layer protocols. This project will study a new Multi-Point Control Protocol (MPCP) in EPON and a new assignment for each ONU a Logic Link Identifier (LLID).



Title : Radio resource management technologies for B3G wireless access


    Principal Investigator: Chung-Ju Chang

Sponsor: National Science Council

    rdKeywords: Beyond 3 Generation (B3G), Heterogeneous Radio Networks, Radio

    Resource Management

    One practical approach of building the radio access networks of the future (beyond rd3 generation, B3G) is to use an integrated heterogeneous network architecture, where radio access networks that are and/or to be designed for specific application

    rdenvironments, for example WLAN and/or 3 generation mobile cellular systems are

    integrated to an all IP core network. Equipped with a properly designed interface and protocols along with joint radio resource management (RRM) over all the participating radio networks, this integrated heterogeneous system can provide users with seamless multi-media Internet services anytime, anywhere at the cost affordable to most of the ordinary users. One of the key enabling technologies for the success of this future heterogeneous radio access system is the development of effective RRM so as to increase the overall spectral efficiency and bring the services cost down. In this joint project, four faculty members with complementary expertise in the Department of Communication Engineering in National Chiao Tung University are brought together for the development of effective RRM for the B3G heterogeneous network system. This joint project is set to accomplish the following main objectives in three years.

    A. Development of a practical, highly flexible simulation platform for RRM in

    a heterogeneous networks environment: By employing site-specific propagation

    models along with realistic mobility model, traffic model, multipath fading and

    correlated shadowing, this platform provides a realistic tool for the study of

    practical systems. Semi-analytical methods will be applied whenever possible to

    have a good trade-off between accuracy and simulation time. This platform is also

    very flexible so that different access systems, such as TDMA/FDMA,

    CDMA/FDD, CDMA/TDD, OFDM-based systems can all be accommodated in

    the simulations. In addition, the platform will be developed under the framework

    of Ns-2 (network simulator 2), which is an open-source simulation framework,

    and therefore, it is easily accessible to others.

    B. Development of new effective RRM algorithms for B3G heterogeneous

    network systems: New effective RRM algorithms such as power control,

    handover, admission control, load control, packet scheduling, dynamical channel

    assignment and mobility management will be developed for B3G systems.

    Specifically, new RRM methods will be developed for heterogeneous networks in

    subproject-1, for packet switched systems in subproject-2, for OFDM-based

    multiple access systems in subproject-3 and new mobility management methods

    will be developed in subproject-4. In addition to be published in technical journal,

    our new results will also be filing patents.



Title : Radio resource management technologies for heterogeneous access


    Principal Investigator: Chung-Ju Chang

    Sponsor: National Science Council

    Keywords: Heterogeneous access network, WCDMA FDD, WCDMA TDD, WLAN,

    Call admission control, Power control, Dynamic channel allocation and


    The future trend of wireless communication system is to provide personal, broadband, and high mobility wireless multimedia services. The mobile users service

    requirements and terminal capabilities diverse such that it is hard to fulfill all the goals within a single communication system. Here comes the heterogeneous access

    network. It is an integrated system that consists of PAN, WLAN, and cellular network, and can provide seamless service transmission according to users location,

    mobility, rate requirements and service QoS requirements. Since the wireless spectrum is scarce, it is necessary to manage efficiently the system resource, including: spectrum, transmission power, access control and etc. In the 3G WCDMA system, within a single core network, there is a plan to support multiple radio access technologies, including: FDD mode, TDD mode and even HiperLAN/2. In this project, the heterogeneous access network that integrates WCDMA FDD/TDD or WCDMA /WLAN is considered.

    To provides wireless multimedia services with heterogeneous service types, rate requirements, and QoS requirements, and also effectively predict the traffic varying of the integrated services, a call admission control algorithm is studied to manage the system resource while the QoS of all the connections can be fulfilled and the spectrum efficiency is optimized.

    In the WCDMA system, the code channels are not exactly mutual orthogonal and the signal mutual interference occurs; therefore, the power control algorithm play an important role in the WCDMA system resource management. Since the transmission characteristics of FDD and TDD modes are different, the mutual interference problem may worsen. In this project, the performance of the power control algorithm in the

    WCDMA FDD/TDD heterogeneous access network is studied. Also, the power control algorithm combing with advanced transmission/receiving technologies (such as MUD and beamforming) is studied. The joint operation can further improve the system resource efficiency.

    Asymmetric transmission is another important characteristic of future wireless multimedia services. In WCDMA TDD system, the up-link/down-link transmission rate can be adaptively changed according to different conditions. Therefore, it fits the requirements of asymmetric traffic. TDD system adopts TD-WCDMA multiple access technology. The changing of the channel allocation will also place an impact on the radio resource management, power control, inter-cell channel allocation such that dynamic channel allocation (DCA) is a very important issue in the radio resource management. In this project, the design of adequate switching point algorithm is studied such that the switching point can be varied according to the up-link/down-link traffic load.

    The handoff algorithm can improve the system capacity and maintain the service continuity in the WCDMA systems. In the WCDMA FDD/TDD heterogeneous

    network, the MAI interference during handoff transition period will vary rapidly such that the packet transmission quality degrades. In the project, the MAI interference model and dynamic guard channel algorithm are studied. Also, in the

    WCDMA/WLAN heterogeneous network, the bandwidth shortage problem may occur during handoff transition period. In this project, the users satisfaction index

    (USI) for wireless multimedia services and the resource allocation for handoff calls are studied.



Title : Speech Training Aids for Hearing-Impaired Mandarin Speakers(3/3)

    Principal Investigator: Wen-Whei Chang

    Sponsor: National Science Council

    Keywords:Hearing impairedVoice conversionHarmonic sinusoidal analysisBark

    spectrumPrincipal-component analysisPitch contour

    The purpose of this three-year research is to develop a speech-training software that helps the hearing-impaired to learn the pronunciation and lexical tones in Mandarin. The first part of this project will focus on developing a voice converter that modifies misarticulations and prosodic deviations of their impaired speech. The key to

    voice conversion lies in the detection and exploitation of characteristic features that distinguish the impaired speech from the normal speech at segmental and prosodic levels. Segmental features that contribute to speech individuality are encoded in the spectral envelope, whereas prosodic information can be found in pitch, energy, and duration variations that span across segments. In light of the low-redundancy spectral representation, we will partition each syllable into an initial and a final, and then perform principal-component analyses of Bark spectra. The principal components are characterized under the form of a Gaussian mixture model with parameters converted using a mapping function that minimizes the spectral distortion between the impaired and normal speech. Next, we propose a codebook-based approach to prosodic conversion that involves matching the pitch, energy, and duration characteristics of the impaired speech with those of the normal speech. Parameters extracted from the pitch contour by discrete Legendre orthonormal transform will be used in the pitch-scale modification. During the third year, spectral and prosodic conversion techniques will be incorporated into the design of visually based training software. It takes the impaired speech as the input, converts it at both segmental and supra-segmental levels and outputs on the computer screen the acoustic correlates that distinguish the converted from the unmodified speech as the visual cues for self-correction. To avoid training on goals that are perceptually insignificant, variation ranges derived from acoustic analyses will be drawn around the visual representations of the converted speech as the training target.

    NSC93-2213-E009-123 (93R280)


Title : ITS information access using voice over

    Principal Investigator: Wen-Whei Chang

    Sponsor: National Science Council


     The purpose of this three-year research is to develop a spoken dialogue system that allows drivers to use voice-controlled commands to access the ITS information server through a mobile ad-hoc network (MANET). The first part of this project will focus on developing a distributed speech recognition system, in which speech features extracted from a local front-end are transmitted through a data channel to a remote back-end recognition server. In light of the low-bit-rate transmission, speech features are compressed using a split vector quantizer that produces the index of the nearest codevector. However, transmitting indexes over noisy channels changes the encoded

    information and consequently leads to severe degradation in the recognition accuracy. This motivates our research into trying to devise a joint source-channel decoder that exploits both the residual source redundancy and the channel memory information through the use of minimum mean-square error estimation. Our decoder design is based on a finite-state Markov chain model that more closely characterizes the temporal dependency in packet losses. Another important issue to address is the perceived voice quality that is contingent on the playout buffer algorithm and on the particular forward error control (FEC) technique used. At the expense of additional end-to-end delay, playout buffer algorithms at the receiver compensate for jitter and FEC is used to mitigate the impact of packet losses. We will develop a software-based measurement system to measure packet loss, delay and network jitter characteristics of voice over IP in a MANET environment. The results can be used to design an efficient voice quality prediction method for joint optimization of FEC and playout buffer algorithm in voice transport over wired and wireless networks. NSC93-2218-E009-041(93R608)


Title : Adaptive coded modulation and bandwidth management (2/2)

    Principal Investigator: Wen-Thong Chang

    Sponsor: National Science Council

    Keywords:Adaptive coded modulation, Multipath interference, Channel capacity.

    In 802.16, both the single carrier system (SC) and the OFDM system support pro