Algorithm for Allocating Outside Optical Fiber in Optical Access Networks

1. Proposed algorithm Reference algorithm Allocation rate Conventional algorithm Central office ns = 40 p = 0.1 T = 0.5 Number of Branched (2x) fl to next area fd fa Allocation area p,ns Algorithm for Allocating Outside Optical Fiber in Optical Access Networks Ikuya Takahashi, Hiroshi Uno and Koichi Sano NTT Access Network Service Systems Laboratories 1-6 Nakase, Mihama-Ku, Chiba-Shi, Chiba-Ken, 261-0023,JAPAN TEL:+81-43-211-3375 FAX:+81-43-213-0941 E-mail: {iku, uno, sano}@ansl.ntt.co.jp Abstract This work details a new optical fiber allocation algorithm that enables a quick service delivery and facility-effective allocation. An optical access network (OAN) consists of NEs like outside fiber cable, inside fiber cable, and service equipment. The outside optical cable has a fiber ribbon structure consisting of four or eight fibers to reduce cable size, weight, and connection costs. Since a technology is not yet available to cut a single fiber in a fiber ribbon and connect it to a fiber in another ribbon, the fiber ribbon has to be cut to allocate one of the fibers. However, there is no quantitative and clear criteria for cutting fiber ribbon, the operator makes a judgment based on experience, or the system allocates the fiber ribbon in ascending fiber-id order. This leads to delays in service delivery and inefficient use of fiber. Therefore, an auto-allocation algorithm that takes into account all relevant information is needed for a quick service delivery and facility-effective allocation. The basic concept of our approach is to allocate the fiber ribbon least likely to be used in the future by considering the fiber ribbon state throughout the cable. In other words, it is to select the cable with the most spare ribbons and allocate one of them. Spare ribbons mean “cannot be used ribbons.” The number of “cannot be used ribbons” depends on the number of “could be used ribbons.” The number of ribbons that will be needed in a certain allocation area (= “could be used” number) can be calculated from the number of subscribers and the probability of demand in the area. Since demand numbers fluctuate, a “safety rate” needs to be defined. The “safety rate” gives the upper limit of cumulative probability of demand and determines numbers of “could be used ribbons.” The number of spare ribbons is counted by subtracting number of actually used fiber ribbons (or could be used fiber ribbons, if it is larger than actually used ribbons) from number of available end fiber ribbons (Fig. 1). The number of spare ribbons in target area is counted by adding up all the spare ribbons in the area (Eq. (1)). The target area can be considered a triangle with a vertex cable. To allocate the ribbon least likely to be needed in the future, following steps are executed. (1) divide the triangular area into two triangles, and count the number of spare ribbons in each, (2) then select the area with the most spare ribbons, (3) repeat steps (1) and (2) until last cable, (4) finally, select the unused fiber ribbon in the last cable and allocate one of its fibers at the demand point connected to the fiber. We simulates the proposed algorithm to evaluate the basic characteristic compared to a conventional algorithm and reference algorithm. Figure 2 shows the allocation rates of each algorithm. The allocation rate means the number of allocated fibers normalized by the one derived from conventional algorithm. Simulation shows that the proposed algorithm has an allocation rate 1.0 - 1.6 times better than that of a conventional outside fiber allocation algorithm and more than equivalent to that of a reference one. (1) gives x, the Cumulative probability ft = larger number of actually or could be used fiber ribbons fl = number of end-fiber ribbons fd = number of dead-end fiber ribbons fa = number of allocated fiber ribbons p = probability of fiber demand in allocation area ns = number of subscribers in allocation area T = cumulative probability ( = safety rate) Figure 1. fiber-ribbon state Figure 2. Allocation rates of each algorithm