Key problems and solutions of all-optical network survivability

Due to the emergence of OXC and Oadm, networking can not only be carried out in the business layer, but also become a reality of the network of optical layer. How to control and manage the network becomes a very important and very difficult problem. Since all optical devices are electrically controlled now, how to construct a suitable control platform on top of the light layer becomes the key problem. Now the international four organizations to make a variety of drafts, although there are differences, but the basic ideas are similar. Their general idea is to IP as the core, the use of IP network traffic control (TE), Is-is, OSPF, RIP, RSVP, and a series of protocols, and on the basis of the original agreement to expand and modify to achieve the purpose of controlling optical network, and the Signaling network (control Platform) is separated from the business network (at least logically separated).

Optical layer of networking, after all, different from the electrical layer, and now to do is not only a single granularity of control and scheduling, but based on small to a packet, TDM, ATM services, large to optical wavelength, fiber-level multi-granularity scheduling and distribution. So the control platform, or the construction of the signaling network, is not only related to a level of problems, but involves several levels, both forward compatible and backward compatible, both to ensure that the optical network of large capacity, can be smoothly upgraded expansion and robustness, but also to ensure the business of QoS and Cos. To achieve such a powerful, integrated signaling network, there are many issues that deserve careful study.

Fault detection and positioning

The major goal of fault detection and localization is to realize protection switching and business recovery quickly and accurately, followed by the whole network management and maintenance.

Since so far, the parameters that can be accurately detected on the optical field are only the optical power, the signal-to-noise ratio (OSNR) and the central wavelength, while the parameters such as BER, LOS and LOF can only be detected in the domain. Because the fault detection is the first step to achieve self-healing, and the protection switch and the recovery of the route have strict time requirements, so the fault detection not only requires accurate, but also requires fast.

Fault location is the second step to realize protection switching and route recovery, and it is a crucial step to realize self-healing, it is necessary to realize the fast and accurate location of fault. Because only determine the specific location of the fault, in order to protect and restore the network and further business configuration, if a network can not be accurate fault location, then the survival of the network will not be guaranteed.

However, due to the transmission of fault, the imperfection of detection mechanism and the time difference between fault detection and fault transmission, the location of fault is difficult in optical network. When fault occurs in a network, such as fiber breakage and node power off, related nodes and detection points have to produce alarms, that is, the situation is often a fault, the network generated alarms, so must know the whole network of information (including configuration and business information), in order to accurately positioning. Of course, this is not insurmountable difficulties, the location of the fault in addition to the function of the node itself closely related to the transmission of signaling or signaling network total control methods directly related. In general, the exact location of the fault can only be achieved if the information of the whole network is known. However, due to the occurrence of the fault is contingency, that is, the information received by each node is abrupt, can not take the general Markov theory analysis, that is not a general m/m/x queuing model, but belongs to the d/m/x model, If each node (distributed control) or the master node (centralized control) needs to collect all the fault information to handle it, so not only a set of complex algorithm, but also the node processing the information must be reasonable consideration of the fault transmission and information transmission time difference, this increases the complexity of implementation, and it is difficult to meet the timeliness of the requirements.

Solution: For distributed control networks, the distributed location method is used, and the distributed fault location approach is related to business and link. For centralized-control networks, that is, the network with the main control node, fault location using a hierarchical positioning method, that is, first of all, a single node to assume a part of the fault of the simple analysis, such as: This node-related alarm filtering and shielding, as well as the node critical alarm positioning, and then the processing results reported to the By the main control node further analysis and processing, and finally determine the exact location of the fault.

From the above analysis, it can be seen that the single node can locate the fault is limited, all the relevant alarms must be collected for correlation analysis, in order to achieve accurate positioning of the fault. This relates to the issue of how to properly and correctly publish the information after single node processing. This is related to the fault information coding, reasonable and effective coding will make the fault location really fast and accurate. Of course, when the fault information is released, it must be supplemented with the corresponding fault information suppression, otherwise, the continuous release of the fault information will cause the fault location and joint errors.