| Summary: | 博士 === 國立清華大學 === 資訊工程學系 === 95 === In this research, we study how to reduce packet delays in single-hop WDM networks with reservation-based packet scheduling algorithms. In reservation-based packet scheduling algorithms for single-hop WDM networks, one of the wavelengths is used as a control channel which is shared by all of the nodes for transmitting reservation information. Time on the control channel is divided into control frames of fixed length. Each of the control frames is further divided into a number of control slots. Before transmitting a data packet, a node sends a control packet in a control slot to inform all other nodes of its intention to send a data packet. The data packet can then be scheduled for transmission.
Packet delay has been the major concern in designing reservation based scheduling algorithms. A packet delay is divided into two major parts, namely, control delay and scheduling/delivery. The control delay starts from the time instant when the data packet arrives at the source node until the time instance when the associated control packet is received by all of the nodes. The scheduling/delivery delay starts from the time instance when the control packet is received by all of the nodes until the time instance when the data packet is received at the destination node. In this research, we are going to study how to reduce control delay and scheduling/delivery delay in single-hop WDM networks.
To reduce the scheduling/delivery delay, two approaches are used. The first approach is to develop a new reservation-based multicast scheduling algorithm. In previous research \cite{bore95a}, the author makes the reservations for data packets starting from the time beyond which the receivers of all of the destination nodes are free. However, between the current time and the time when the receivers of all of the destination nodes are free, there could be a number of free data slots on some data channels which can be used to schedule data packets. We propose to make use of every data slot on all data channels such that the performance of the system can be improved.
The second approach for reducing the scheduling/delivery delay is to replace the tunable trans-ceivers in all of the nodes with fixed transceiver arrays such that no tuning latency is required. Furthermore, each node is capable of transmitting and receiving multiple signals on multiple channels at the same time. This approach enables the network to have better tolerance to traffics with non-uniform distribution of destinations that are commonly seen in network traffics.
To reduce the control delay, three approaches are proposed. The first approach is designed for tunable transceiver based single-hop WDM networks. The second and the thrid approaches are designed for transceiver array based single-hop WDM networks.
Our first approach aims to increase the utilization of the control slots and reduce the packet delays in tunable transceiver based single-hop WDM networks. We propose to adjust the lengths of the control frames dynamically according to the traffic patterns of the nodes. The structure of dynamic control frame (DCF) has variable number of control slots. A scheme is devised to enable the nodes in the network to dynamically acquire and release control slots depending on their loads. Thus, the control frames will include only necessary control slots for those nodes that have or potentially have data packets to send. Therefore, the control slots on the control channel can be used efficiently and the packet delays can be reduced.
With a fixed transceiver array in each of the nodes, each channel becomes an independent broadcast channel such that broadcast and multicast communications can be efficiently supported. With fixed total number of channels in a transceiver array based single-hop WDM network, there is a tradeoff between the number of channels used for transmitting reservation (or control) information and the number of channels used for transmitting data packets. A network with fixed numbers of control and data channels is not able to cope with different traffic patterns. The optimal combination of the numbers of control and data channels depends on the traffic patterns.
Our second approach for reducing the control delay is to develop an adaptive channel allocation algorithm to dynamically adjust the numbers of control and data channels according to the traffic patterns. The proposed adaptive channel allocation algorithm enables a transceiver array based single-hop WDM network to dynamically choose the best combination of the numbers of control and data channels for wide ranges of traffic patterns including non-uniform arrivals, non-uniform destination distributions, broadcast/multicast communications, and different system loads.
Our third approach for reducing the control delay is to use DCF over transceiver array based single-hop WDM networks. However, the proposed DCF scheme introduces extra overhead in each control frame resulting in higher mean packet delay than using fixed control frame (FCF) structure when the load of the system is heavy and the degree of non-uniformity of traffic patterns is low. Thus, we propose a hybrid fixed and dynamic scheme that employs the FCF scheme or the DCF scheme depending on the required number of control slots in each control frame. The hybrid scheme is able to choose the better of the FCF and DCF schemes such that it yields the lowest mean packet delay for the entire range of system load and degree of non-uniformity of traffic patterns.
In this research, we thoroughly investigate the factors that affect the packet delay of reservation-based packet scheduling algorithms in single-hop WDM networks. Based on the understanding of the factors that affect the packet delays, several algorithms are proposed to reduce control and scheduling/delivery delays. Our simulation results show that the proposed algorithms are able to significantly reduce the average packet delays of single-hop WDM networks.
|
|---|
