Recent improvements in the performance of personal computers and workstations have caused the need for communication networks that provide greater functionality and bandwidth. High-performance local-area-networks have developed into communication infrastructures essential for efficient business operations. Furthermore, communication networks are expected to progress from voice-only telephone communication networks to broadband integrated services digital networks (B-ISDN). Network survivability is one of the key issues in realizing B-ISDN. Since the B-ISDN handles a large amount of information, a network failure will paralyse social and business activities. For this reason the future B-ISDN will require high survivability and consolidate management capabilities for reliability.
Synchronous optical network/synchronous digital hierarchy (SONET/SDH) is the most suitable technology for satisfying these demands, because the SONET has an advanced operation and maintenance (OAM) function. A new network application, which has excellent survivability characteristics is emerging; it is called a self-healing ring.
A self-healing ring consists of add-drop multiplexers (ADM's), and it guards against fiber cuts without using the more expensive technique of one-for-one (1:1) protection switching with diverse routing. It also reduces the number of nodes required for the same traffic by allowing fiber capacity to be shared. Using these rings, thus improves network survivability and availability while reducing cost. Hence SONET self-healing rings are expected to form the major network infrastructure in future B-ISDN.
The self-healing ring can generally be divided into two categories:bidirectional SHR's (B-SHR's) and unidirectional SHR's (U-SHR's). The type of ring depends upon the path travelled by a duplex communication channel between each office pairs. The SHR is called a bidirectional SHR (B-SHR) if both directions of a duplex channel travel over the same path; a unidirectional SHR (U-SHR) if the direction of a duplex channel travel the opposite paths. For examples in fig 1(a) and (b), both directions of a duplex channel between offices 2 and 4 use the same path which travels through office 3. For the case of U-SHR (fig 1(c)) a duplex channel between offices 2 and 4 travels over two opposite paths: path 1: 2->3->4, and path 2: 4->1->2. Thus a B-SHR requires two working fibers to carry a duplex channel, and a U-SHR requires only one working fiber to carry a duplex channel. In order to provide a protection capability for fiber system failures and fiber cable cuts, a B-SHR may use four fibers (one working fiber pair and one protection fiber pair) or two fibers (all working fibers with the spare capacity for protection), and a U-SHR requires only two fibers (one working fiber and one protection fiber).
Figure 1 :SONET self-healing rings. (a) Bidirectional SHR with 4 fibers (B-SHR/4). (b) Bidirectional SHR with 2 fibers (B-SHR/2). (c) Unidirectional SHR (U-SHR)
For each type of rings, two possible SONET self-healing control schemes may be used: line protection switching and path protection switching. The line protection switching schemes uses SONET line overhead for protection switching and restores line demand from a failed facility, while the path protection switching schemes uses SONET path overhead, and restores individual end-to-end service channel. Today, only unidirectional rings with path protection switching and bidirectional with line protection switching are commercially available in the U.S market place. So, a unidirectional self-healing ring (USHR)mentioned here is referred to as a unidirectional, path-switched ring. Similarly, a bidirectional self-healing ring (BSHR) is referred to as a bidirectional, line-switched ring.
1) Bidirectional Ring with Four Fibers (B-SHR/4): This architecture has evolved from a majority of today's asynchronous point-to point systems (or hubbed protection systems). The protection capability of this B-SHR/4 architecture is achieved by using automatic protection switching systems (APS's) to perform a loop back function in case of cable cuts or node failures. For example in fig. 1(a), working traffic in the B-SHR/4 travels bidirectionally on separate fibers of the same path and dual protection fibers serve as back-up. In the case of a cable cut, traffic is intercepted at the next CO and re-routed back to its destination on the protection fibers. The B-SHR/4 architecture requires a protection ADM for each working ADM and 1:1 nonrevertive lower-speed electronic protection switch at each office. The nonrevertive 1:1 switch is a protection switch such that the signal need not be switched back when the failed line is repaired.
2) Bidirectional Ring with Two Fibers (B-SHR/2): For this scheme the traffic is routed on a two-fiber ring with one ADM at each office. In the normal situation, traffic is evenly split into the outer ring and the inner ring by filling even and odd numbers of time slots (fig. 1(b)). If there is a fiber break or equipment failure, traffic is automatically switched into vacant time slots in the opposite direction to avoid the fault.
The U-SHR architecture uses only two fibers with one for working and the other for standby and an ADM at each office. For each U-SHR, the directions of a duplex channel travel on different routes between two offices in the SHR. The self-healing capability is achieved by using low-speed path selection. A SONET USHR architecture discussed here is based on a concept of signal dual-feed (1+1 protection). The architecture of such a ring consists of one ADM at each office and a pair of fibers with traffic going to the other direction.
In the economic analysis which follows, we will only consider the equipment (ADM) cost, which is a dominant factor of the total fiber transport cost (cost for terminating equipment, fiber material, and fiber placement) for intra-LATA networks.
1) B-SHR/4 Versus U-SHR: Both the B-SHR/4 and the U-SHR use the dedicated spare capacity that can makes the control system simpler. In other words, the ADM cost for a B-SHR/4 and a U-SHR can be assumed to be approximately the same. Thus comparing the B-SHR/4 with the U-SHR in terms of cost and capacity can fair. In general, the B-SHR/4s have higher capacity than the U-SHR's but at the penalty of additional components and facilities such as more fiber and ADM's. The cost of a B-SHR/4 is approximately twice the cost of a U-SHR at the same rate since the amount of equipment and facilities for a B-SHR/4 is twice the amount for a U-SHR. However, this capacity comparison between U-SHR and a B-SHR highly depends upon the demand requirement pattern and the network design method.
2) B-SHR/2 Versus U-SHR: Working and protection channels for B-SHR/2 are routed on the same fiber. In the case of network system failures, protection switching is accomplished by using the time slot interchange method. In order to simplify the ADM system design complexity and minimise impact on network operations, the fiber system dedicates half of bandwidth for protection. Since the B-SHR/2 only uses half of the bandwidth, the load balancing (demand splitting) traffic arrangement is usually used to increase ring utilisation. When comparing B-SHR/2s with U-SHR's under the condition that no demand splitting is allowed, we arrive at the same results when comparing B-SHR/4 with U-SHR. Even if we take into account the demand splitting factor for the B-SHR/2, we still find that the relative cost comparison depends strongly on the demand pattern.
1) U-SHR:In the U-SHR architecture two copies of signal are sent around the ring in opposite directions. The destination monitors the quality of the signals received and selects the better one (Fig. 2). Each link of a bidirectional link between two nodes travels the same direction around the ring.
Figure 2: U-SHR architecture.
2) B-SHR/2: In a 2-fiber B-SHR, capacity in each fiber is divided in two: Half for working traffic and half for protection traffic. In a normal state (Fig. 3(a)), communication is conducted using only the working capacity.
Figure 3: 2-Fiber BSHR operation.
When a failure occurs, the protection capacity is used for rerouting the affected traffic. An automatic protection switching (APS) controller triggers a protection switching operation called "ring switched". This operation is performed by by the two ADM's on either side of the failed segment. Both ADM's are placed in "ring switch" ; they then reroute traffic away from the failed segment by using the protection capacity (Fig. 3(b)). They bridge the STS-1 traffic being sent towards the failed fiber over to the protection capacity and switched traffic coming from the protection capacity over to the working capacity. ADM's on the protection path are placed in "full pass-through (FPT)" state; they then connect the input and the output channels.
Fig. 4 shows the internal operation of ADM in fig. 3 for both normal and ring switched states. A TSI function is used in the ADM to achieve these switching states. The FPT state can easily be achieved by the TSI function.
Figure 4: Internal operation of ADM in 2-fiber B-SHR.
3) B-SHR/4:In a 4-fiber network (fig. 5(a)), the working and protection traffic travel on separate fibers. It is not only equipped with the ring switch, but also with a span switch, which is similar to 1:1 linear protection switching. If a failure occurs on a working fiber, the ADM's on either side of the failure are placed in "span switch" state; they then switch the traffic from a failed fiber to a protection fiber on the same span (Fig. 5(b)).
Figure 5: 4-fiber B-SHR operation (a) Normal condition. (b) Span switch operation. (c) Ring switch operation.
If a failure occurs on both working and protection fibers on the same span, a ring switch is triggered instead of a span switch (Fig. 5(c)). The ADM's on either side of the failed segment reroute the affected traffic in the opposite direction away from the failure. The ADM's on the protection path are place on the "FPT" state.
Fig. 6 shows the internal operation of a ADM A in Fig. 5 for each state: normal, span switch and ring switch state. These states and the FPT state can easily be achieved by a TSI function at the STS-1 level. The throughput of a switch in a B-SHR/4 network is larger than that in in a B-SHR/2.
Figure 6: Internal operation of ADM in a 4-fiber B-SHR.
A U-SHR is the simplest ring architecture. Since the U-SHR does not require any communications between nodes for protection switching, it is easily realized. However, it cannot use the bandwidth efficiently. The bandwidth requirement of the U-SHR is the maximum bandwidth requirement over any span between any two nodes in the ring. No reuse of bandwidth is possible in a U-SHR network. Furthermore, it is impossible to transmit extra traffic, which is traffic transmitted on protection capacity only under normal conditions, because the protection capacity is always used for the protection switching operation in a U-SHR network. The architecture achieves its maximum efficiency when all traffic on the ring is between a hub node and other remote nodes i.e it is more suitable for access networks.
In the B-SHR architecture, the shorter path is used for normal communication and the longer path is used for back-up. If one fiber goes down, the ring sends traffic in the other direction. Communications between the two nodes adjacent to the failure achieves this rerouting operation. The internodal communication channel is accomplished via messages over the automatic protection (APS) channel. Since a B-SHR needs communications between nodes, a protection switching operation of a B-SHR is more complex than that of a U-SHR.
Though B-SHR architecture is complex in its operation , it has the advantage of maximising bandwidth utilisation and has a higher capacity because the B-SHR provides the ability to reuse bandwidth and support extra traffic on the protection capacity. Thus , due to its bandwidth efficiency, the best application of the B-SHR architecture is for transport network.
|||T-H Wu, "Emerging technologies for fiber network survivability", IEEE
Communications Magazine, Feb 1995.
|||T-h and R. C. Lau, "A Class of Self-Healing Ring Architectures
for SONET Network applications", IEEE '90
|||M. Tkatori, "A High Performance Switch for OC-12 SONET
Self-Healing Ring Networks" IEE, Feb. 1996