Kai-Wei Ke

The design of QoS provisioning mechanisms for wireless networks

Two mechanisms are proposed in this paper to be incorporated with IEEE 802.11e EDCA MAC protocol for providing the QoS provisioning capability of such network. First, the call admission control function, which considers the inherent contention access feature of 802.11 MAC, aims to satisfy the QoS requirement of established higher priority connections, while protecting the minimum reserved bandwidth of traffic flows with lower priority at the same time. In addition, the proposed adaptive contention window adjustment mechanism dynamically adjusts the maximum and minimum contention window size based upon the equivalent established connection number observed by each individual access category. Such approach reduces the channel contention probability and packet access delay effectively. Performance results via simulations have shown the advantages achieved by employing these two mechanisms.

Mobile awareness based cluster selection mechanisms in wireless ad hoc networks

We propose an efficient clustering scheme for a wireless ad hoc network, based on the moving characteristics of the mobile nodes. The instantaneous mobility behavior for a node equipped with a mobility measuring tool, such as GPS, is readily available. However, for a mobile node without such tool or one in an environment where the measuring tool cannot function properly, we develop a simple heuristic mechanism to derive the mobility index, used for characterizing its instantaneous nodal mobility. The developed mobility estimation algorithm is fully distributed and self organized. Performance results via simulations reveal that the estimated mobility index of a node follows a linear relation with its own real speed in various networking environments. Next, we apply the obtained mobility index to derive the mobile awareness based cluster selection scheme. In general, we choose the currently slowest moving node as the primary candidate for the clusterhead. Compared with other well known clustering schemes, performance results have demonstrated that our mechanism can achieve a more efficient and stable clustering architecture for different networking topologies.

State reduction and call admission in multirate circuit-switched networks

In a circuit (real, or virtual) switching network, call admission control plays a central role in determining the throughput and grade of service of a network. This issue, of course, is not new, but the characteristics of multirate and high speed of a broadband integrated service digital network (ISDN) make the problem quite different from its counterpart in a telephone (single-rate) network. The tremendous number of states, created by the huge bandwidth of a broadband network and its multirate characteristic, makes the design and analysis of call admission policies in a broadband network computationally intractable. We propose a state grouping method to reduce the total number of states. An important characteristic of the method is that the total gain of a policy remains unchanged in the reduced state space. This means that the search for a better policy can be conducted in the reduced state space created by the state reduction method. The time complexity is drastically reduced.

Quantization and cost computation of MDP-based admission and routing

Quantization, a process to restrict a continuous variable to a discrete one, has been a powerful technique for many fields of engineering. We show how to use the concept of quantization to compute the cost function of a Markov decision process (MDP)-based routing and admission policy in a broadband network. In the discussion we assume a multirate circuit switching environment.

A switched delay line based optical switch architecture with a bypass line

This paper proposes a new switching architecture to be used in all-optical networks. The proposed switch, M-B-Quadro switch, is extended from an original 2/spl times/2 two-stage multi-buffer switched delay line based optical switching node, known as M-Quadro node. By incorporating bypass lines and employing a novel switch control strategy, denoted as LAVS, the switching node can effectively resolve packet contentions, thus reducing the packet deflection probability substantially.

THE DESIGNS OF A SCALABLE OPTICAL PACKET SWITCHING ARCHITECTURE

Abstract This paper proposes a new switching architecture to be used in all optical packet switching networks. The proposed switch is derived from an original 2 × ?2 two‐stage multi‐buffer switched delay line based optical switching node, known as an M‐Quadro node. By incorporating bypass lines into the M‐Quadro architecture and employing a novel switch control strategy, the optical packet switching node can effectively resolve packet contentions, thus reducing the packet deflection probability substantially. Furthermore, we show that such architecture is scalable for a generic multiple stages optical packet switch with a larger number of input/output ports.

A Novel Mobility Model of Diverse Average Speed and Uniform Node Distribution in Mobile Ad Hoc Networks

Most wireless protocols and algorithms proposed in mobile ad hoc networks (MANET) are evaluated and compared by using simulations. The impacts of various simulation parameters are noticeable. However, the effects of the mobility model do not attract many attentions until recently. The random waypoint (RWP) mobility model, due to its simplicity, is one of the most often deployed mobility models in simulations. But it implicitly introduces many defects such as average speed decaying and un-uniform node distribution. In this paper, we propose a novel mobility model, general ripple mobility model (GRMM), to reduce the above mentioned problems in RWP. Furthermore, GRMM can provide another great feature, diverse average speed (DAS), to generate different average speed mobility patterns easily within the same speed range. Simulation results have shown that GRMM can simultaneously achieve uniform node distribution and diverse average speed

Quantization, state reduction and cost-based admission and routing

Cost-based admission and routing have been widely discussed and used in telephone switching networks, but their application in a multirate high-speed network is severely hindered by the enormous number of states of the Markov model describing the network. In this paper we show how to apply the concept of quantization to reduce the states of the model. Once the state space is reduced, conventional techniques can then be used to compute the cost function for a cost-based admission and routing scheme. The proposed method is policy-independent, meaning that it can be applied to different call admission policies.

Performance evaluation of multisource Application Layer Multicast (ALM): Theoretical and simulative aspects

Multisource ALM routing protocol is an indispensable mechanism for efficient distribution of data from many sources to many receivers. The growing of Internet multimedia applications (e.g. videoconference and multiparty online game) which demand large network resources and call for multisource multicasting, has explicitly developed the necessity of an effective method to evaluate and compare the efficiency of various multisource multicast routings. We define a routing cost to be the total number of underlying hops traversed by multicast data over a multicast delivery tree. This study explores a multiplicative property to obtain the minimum cost of various multisource ALM routing protocols by simply multiplying the number of sources to the minimum cost in the single-source multicast. From theoretical and simulative studies, the multiplicative property has been proven through examining the minimum cost of multisource multicast delivery trees constructed by different multisource ALM routing protocols with source-specific tree (SST), core-based group-shared tree (GST-C) and bidirectional group-shared tree (GST-B), which are employed in the design of multisource ALM routing protocols.

Performance analysis for hierarchical resource allocation in multiplexed mobile packet data networks

The general packet radio service (GPRS) can be thought of as an overlay network above a Global Systems for Mobile (GSM) network. In cooperation with the GSM system, GPRS supports both voice and packet data services. Obviously, combining voice and data traffic implies the nature of contending with finite radio resources. The paper focuses on voice/data integrated GPRS downlink traffic and proposes two hierarchical resource allocation strategies by considering time slots and radio blocks simultaneously with two different TDMA frame configurations. The radio block based resource allocation can characterize the multiplexing scheme of the practical GPRS operation. The performance of the proposed strategies were evaluated analytically by multidimensional Markov chain and verified through intensive computer simulations. Based on both the analytical and simulation results, it showed that the proposed allocation strategies can offer high utilization of transmission resource, as well as low blocking probability to GPRS users. The results of the study provide not only a practical thinking for implementation but also a guideline for analysis.