No abstract

In a reconfiguration mechanism for improving the RAS of a ring LAN, it is necessary to guarantee the proper functioning of communication functions up to the medium access control (MAC) layer. This paper proposes a reconfiguration mechanism for distributed-control dual ring LANs which maximizes the size of the ring while maintaining MAC layer communication functions by automatic ring merging and wrapback. Each station bases its actions on the state of the two input links. When a link fails, the failed segment is isolated by ring wrapback and the ring operates in wrapback mode. When multiple failures occur, the ring is wrapped back further into a smaller ring or several rings until the MAC layer functions are recovered. On the other hand, when the MAC layer functions of the failed segment are recovered, wrapback is released and the failed segment is merged into the active ring. With this reconfiguration mechanism, the largest ring possible to maintain MAC layer functions is formed automatically regardless of failures, powering up or down, etc. This mechanism is represented using finite state machine (FSM) diagrams. The recovery times from multiple failures are evaluated using the specifications of the international LAN standard (IS0 8802-5). This reconfiguration mechanism has been approved as the IEEE standard (recommended practice) for a dual-ring reconfiguration mechanism, and its effectiveness has been proven by field trials.

In the distributed‐control ring LAN (local area networks), a reconfiguration control is required to cope with the independent asynchronous failure detection in the stations. This paper proposes a wrapback reconfiguration mechanism in the dual‐token ring LAN with a parallel dual structure of distributed‐control token‐ring LAN. Here, the ring is wrapped back automatically by the distributed‐control of the stations when a fault is detected in the communication function in the physical layer or MAC (medium access control) layer. Each station connected by the dual ring determines its operation based only on the state of the two input links. When an abnormality is detected in a link, a failure‐detecting beacon is transmitted and the failure domain is located by the reception priority competition control, where the transmission is stopped by receiving the beacon. The stations adjacent to the failure domain artificially generate a fault in the output link paired to the failure link, so that the station to wrap back the ring at both sides of the failure link‐pair can be located by the same procedure as for the localization of the failure domain. In parallel to the ring wrapback, the active monitor is established, which is needed in the initialization of the ring. Each station discriminates between the current link and standby link, and the ring wrapback by the failure of the standby link is avoided. Applying the international standard (IEEE 802.5) to the token‐ring LAN, the failure recovery time is evaluated. The effectiveness of the proposed reconfiguration mechanism has already been verified by an experiment using the actual equipment.

The in-house information communication network using local area network (LAN) is required to cope flexibly with the expansion/modification of the system and to avoid the systemdown in the generation of a fault by detaching the fault domain. This paper considers the star-ring LAN with the ring spurs and a dual ring. It proposes also the structure and control of the wiring concentrator of ring LAN with the dedicated automatic reconfiguration control by the ring spur bypass and the ring wrapback. Since the ring contains both the ordinary terminals and WC, the reconfigurationdedicated medium access control (MAC) frame is introduced anew and can be sent only by wiring concentration (WC), making unnecessary the specification modification of the MAC layer of the terminal.The WC which handles the localization and detachment of the fault point is determined by this mechanism. The test ring is automatically expanded from the local ring in WC to the ring containing the ring spur and the 23 ring containing the dual ring successively by testing the conductivity.Using WC, a ring LAN system with the automatic ring degeneration/expansion function can be realized, while providing the terminal with the function of connection/detachment/shift on LAN (hot-pluggability). The time required for reconfiguration by the proposed reconfiguration control is evaluated by summing up the times for the major control steps and by experimental examination. system, i.e., modification of the network configuration due to the increase/transfer of the terminals. By connecting the terminals in a star topology through the wiring concentrator (WC), it is possible to cope with the modification of the wiring layout due to the connecnection/detachment of the terminals only by using WC. Then a flexible in-house information communication network can be constructed with a concentrated and highly efficient operation and maintenance [ 1, 21.The ring LAN, on the other hand, has relatively less constraint concerning the communication speed and the communication distance and often is used as a typical topology in the in-house information communication network from the viewpoint of the high-speed and largecapacity transmission using the optical fiber and the flexibility in the system configuration [3].From such a viewpoint, the star-ring LAN is considered where ring LAN are connected in a star topology through WC.