Storage Area Network (SAN) has gradually developed as the demand for storage capacity and fast access has increased. The traditional way of attaching storage directly to the servers over a SCSI bus has limited scalability. Several drawbacks and limitations have turned up. Switched Fibre Channel SAN resolves all of these issues. In this paper, the architecture of the switch fabric for the SAN is discussed. The complete design of the free-space optical switching core based on the diffractive element and the PLZT shutter is proposed.
Passive optical interconnection is being used in large packet switches to allow size scale up and more efficient heat dissipation from the electronic processors. The line card is the electronic island whose size is determined by thermal dissipation technology. We have selected an optical interconnection technology that is low cost, fiber ribbon, and we are working on the shutter arrays, which allow active routing. This will reduce the demands on the electronic processors. Progress on a 180 Gbps switch will be reported.
Optical switches based on liquid crystal SLM (Spatial Light Modulators) have traditionally been considered unsuitable for packet switching due to slow reconfiguration speed. In this paper we investigate the constraint of reconfiguration time in an optically interconnected packet switch. A system architecture based on the established knockout principle and input/output buffers is simulated with self-similar traffic patterns and packet length statistics obtained from NLANR. Analysis includes packet delay distribution, queue length growth. A physical realisation of the system will use VCSEL arrays, detector arrays and multi-mode ribbon fibre. Data granularity of the system is chosen to match the specification of modern line cards used in routers. It is found that a reconfiguration time in the order of micro seconds is sufficient for an acceptable delay and loss rate. Relationships between required reconfiguration time and system parameters are established.
This research mainly focuses on the performance evaluation and the improvement of the IEEE 802.11 DCF. The basic access method in the IEEE 802.11 MAC protocol is the distributed coordination function (DCF), which is a carrier sense multiple access with collision avoidance (CSMA/CA) MAC protocol. In this paper, we first evaluate the performance of DCF using the classical Markov analytical model proposed by Bianchi, and then present an improvement named TDCF. Both the theoretical computation and the simulation in our research prove that TDCF has better performance than DCF.
This paper evaluates the performances of the contention-based channel access mechanism of IEEE 802.11e, called enhanced distributed coordination function (EDCF), compared with the 802.11 legacy MAC in supporting voice, video and data applications through network simulation of a scenario of 802.11e. Then we discuss the effects of Contention Window (CW) and Arbitration Inter-Frame Space (AIFS) on service differentiation and total throughput. We also consider an optional feature of the EDCF, called contention-free burst (CFB). Through our simulation study, we can draw a conclusion that the EDCF with TXOP can provide better-differentiated channel access for different traffic types than EDCF without TXOP especially at high traffic load conditions. But the movements caused by the parameters in CFB seem a lot bouncing and instability when in different application and configuration.