In this paper, we investigate packet loss and system dimensioning of feedback (FB) type wavelength division multiplexing (WDM) optical routers under asynchronous and variable packet length self-similar traffic. We first study the packet loss performance for two different types of WDM optical routers under asynchronous and variable packet length self-similar traffic. Based on simulation results, we demonstrate that a 1616 FB type WDM optical router employing more than 4 re-circulated ports without using void filling (VF) algorithm has better performance. We then present the system dimensioning issues of FB type WDM optical routers, by showing the performance of FB type WDM optical routers as a function of the number of re-circulated ports, buffer depth, re-circulation limit, basic delay unit in the fiber delay line optical buffers and traffic characteristics. The sensitivity of the mutual effects of the above parameters on packet loss is investigated in details. Based on our results, we conclude that the FB type WDM optical routers must be dimensioned with the appropriate number of re-circulated ports, re-circulation limits, buffer depth, and optimal basic delay unit in the fiber delay line optical buffers under relevant traffic characteristics to achieve high switching performance.

Broadband ISDN, using asynchronous transfer mode, are expected to carry traffic of different classes, each with its own set of traffic characteristics and performance requirements. To achieve the quality of service in ATM networks, a suitable buffer management scheme is needed. In this paper, we propose a buffer management scheme using a priority service discipline to improve the delay time of delay-sensitive class and the packet loss ratio of loss-sensitive class. The proposed priority scheme requires simple buffer management logic and minor processing overhead. We also analyze the delay time and the packet loss ratio for each class of service. The results indicate that the required buffer size of the proposed priority scheme is reduced and the delay time of each class of service is controlled by a parameter. If the control parameter is appropriately chosen, the quality of service of each class is improved.

The differential 4-quadrature (D4Q) coherent optical system which does not need absolute phase information is proposed. The input information is encoded in the relative position of the present symbol with respect to the reference frame constructed by the previous 3 symbols. The general theory of the system operation including encoding and decoding is presented. This system does not need to track the fluctuation of the states of polarizations which is essential for most other polarization modulation systems. As an example, the 4-symbol D4Q system is described. And the saddle point approximation is applied to estimate the bit error rate performance. The analytic approximation agrees with the simulation results very well.

The symbol basis side information generated by Viterbi's ratio threshold test technique is proposed to improve the performance of the asynchronous slow-frequency-hopped multiple access system with BFSK signaling in the frequency non-selective fading channel. By properly setting the ratio threshold to produce erasure decisions for the received symbols, the system performances are optimized. The relationship among the hit symbols in a hop duration is exploited by this symbol basis side information to greatly reduce the packet error probability. This packet error rate improvement can be as large as two order of magnitude, compared with perfect hop basis side information systems.

An efficient void filling (VF) algorithm is proposed for wavelength division multiplexing (WDM) optical packet switches (OPSes) handling variable-packet-length self-similar traffic. The computation complexity of the proposed algorithm is extremely low. We further compare the switching performance of the proposed algorithm with that of the conventional one. We demonstrate that the proposed algorithm offers significantly lower computation complexity with adequate performance.

Most of the proposed self-similar traffic models are asymptotic in nature. Hence, they are less effective in queueing-based performance evaluation when the buffer sizes are small. In this paper, we propose a short range dependent (SRD) process modelling by a generalized variance-based Markovian fitting to provide effective queueing-based performance measures when buffer sizes are small. The proposed method is to match the variance of the exact second-order self-similar processes. The fitting procedure determines the related parameters in an exact and straightforward way. The resultant traffic model essentially consists of a superposition of several two-state Markov-modulated Poisson processes (MMPPs) with distinct modulating parameters. We present how well the resultant MMPP could emulate the variance of original self-similar traffic in the range of the specified time scale, and could provide more accurate bounds for the queueing-based performance measures, namely tail probability, mean waiting time and loss probability. Numerical results show that both the second-order statistics and queueing-based performance measures when buffer capacity is small are more accurate than that of the variance-based fitting where the modulating parameters of each superposed two-state MMPP are equal. We then investigate the relationship between time scale and the number of superposed two-state MMPPs. We found that when the performance measures pertaining to larger time scales are not better than that of smaller ones, we need to increase the number of superposed two-state MMPPs to maintain the accurate and reliable queueing-based performance measures. We then conclude from the extensive numerical examples that an exact second-order self-similar traffic can be well represented by the proposed model.

In this paper, we study a work-conserving multicast scheduling with fanout splitting in a switch, which routes incoming packets asynchronously without fragmentation into cells. A new switch architecture is proposed, which distributes the input links to P variable length packet switching fabrics (VPS) with every G input links sharing GR inlets of the VPS. The system performance is analyzed by queueing analysis to express the maximum throughput and packet delay in terms of the system parameters and traffic characteristics. A practical switch design is also proposed to realize almost the same scheduling as the work-conserving one. We have surveyed how the fanout distribution affects the performance of the switch through Fanout Function, which is defined and studied to help the design of a multicast switch. We show how Fanout Function determines the maximum throughput and packet delay. Various fanout distributions are compared. The mixed fanout distribution exhibits better performance while the deterministic fanout can be used as a bound in the design of a multicast switch. We optimize R and P to attain 100% maximum throughput under limited switch complexity. When the mean fanout size is large, we can use less hardware to achieve the optimal performance by using our architecture. The proposed realization of this switch can be implemented easily due to its modular design. It is scalable because distributed output contention resolution and routing are used instead of a central arbitrator. Its performance is verified by simulation. The result matches the theoretical work-conserving scheduling very well.