Previous Article Next Article. References You do not have subscription access to this journal. Cited By You do not have subscription access to this journal. Figures 14 You do not have subscription access to this journal. Tables 4 You do not have subscription access to this journal. Equations 10 You do not have subscription access to this journal. Journal Home. Issues in Progress.
Current Issue. All Issues. Feature Issues. End If 9. If both s r s , d and can provide bandwidth r Then Else End If Note that tidal traffic can potentially result in low bandwidth utilization and weak service capability, as network resources are not allocated properly.
To address this problem, in this study, we first analyze traffic characteristics of access networks and metro networks, and mathematically formulate an onion tidal traffic model OTTM in EONs. Second, based on the traditional routing and spectrum allocation RSA algorithm, which provides end-to-end connection by allocating frequency slot resources in EONs, we propose two algorithms to enhance bandwidth efficiency based on the OTTM model.
All network assumptions are also treated as inputs. Thus, the allocation of resources to satisfy the demand is delayed until the current time reaches the book-ahead of the request. A specific modulation level maps to each path on Line 18 depending on the reachability limits specified in Table 2. The number of frequency-slots that are required to satisfy the request to transfer to each destination node on Line 19 is calculated from Eq.
The largest span of contiguous frequency-slots for the duration of the request is computed by a call to the subroutine on Line If the computation on Line 19 finds that there is not enough spectrum available, then it will try the next time-slot.
After trying whole flexible window and not finding enough available spectrum to support the entire request to transfer to a specified replica, then the next replica will be tried. If there is indeed enough total spectrum capacity and sufficient contiguous frequency to completely support the request, request will be served. We use first-fit spectrum assignment for allocation. All source and destinations are uniformly distributed throughout the network. All other assumptions correspond to those listed in Table 2.
We have explored several categories of algorithmic variants. We have also investigated anycast and unicast requests and proposed balanced routing for anycast algorithm to reduce blocking.
We consider the same window flexibility for starting the transmission that is applied in the DSA algorithm for ISA as our baseline to compare them fairly. In flexible STSD if the request is not able to be served after its book-ahead time, it can stay in the queue in hopes of finding available resources before the delay tolerance expires. In Fig. The next category explored is adding anycast flexibility as another combinatorial option.
In Figs. The availability of terminal flexibility via anycast conclusively has a great impact on lightpath establishment. Blocking is decreased as the number of anycast candidates grows. Figures 6 c and 6 d display the average initial delay of DSA and ISA algorithm with flexible 10 time-slots window when two and three replicas, respectively.
Note that by increasing the number of replicas in the network, initial delay will be longer since initial delay is calculating for successful demands which is higher with more replicas in the network. We can observe that with greater number of replica DSA provides higher improvement in terms of blocking probability compared to ISA in both fixed and flexible scenarios.
The reason for this is increasing the number of replicas improves network resource allocation efficiency. We investigated the dynamic selection of least cost path to trade off between resource availability and number of hops of paths in anycast scenarios.
Figure 8 a illustrates blocking probability of delayed spectrum allocation algorithm with fixed STSD, assuming there are two replicas in the network. Considering Eq. Therefore, the best result is obtained when we choose a data center between available slots in the minimum hop path.
Figures 9 a and 9 b show average initial delay of DSA algorithm with flexible 10 time-slot window considering balanced routing when two and three replicas, respectively. The reason for this comes from consideration of link usage that helps in reducing total blocking and less requirement to postpone sending a request in order to prevent blocking of that request.
Least spectrum usage policy cause the higher number of hops that leads to higher delay as well. The difference between their number of hops is at most 0. Increasing the number of replicas leads to fewer hop counts. Selected path traverses more hops, and use more resources in flexible window STSD. The reason is that in flexible window total blocking decreases and more requests are able to be served. Increasing the total requests in the network increases the number of hops for traversing.
In this work, we have investigated the problem of allocation of unicast and anycast traffic demands using the novel approach of DSA in elastic optical networks with fixed and flexible window STSD with balanced routing. This improvement depends on the number of data centers in the network and increases by increasing the number of data centers.
We can demonstrate that by using DSA delay of transmission increases however by decreasing the size of the flexible window the initial delay decreases. We show that by using a balanced routing approach we can decrease blocking without increasing delay. Also, we reach the best trade-off between minimum shortest hop count and spectrum availability in each path.
Some portions of this work have been published in [ 33 ]. Jinno, H. Takara, B. Kozicki, Y. Tsukishima, Y. Sone, and S. Develder, M. Leenheer, B. Dhoedt, M. Pickavet, D. Colle, F. Turck, and P. IEEE 5 , — Zhao, H. Wymeersch, and E. Lightwave Technol. Jinno, B. Kozicki, H. Takara, A. Watanabe, Y. Sone, T. Tanaka, and A. Wan, N. Hua, and X. Zhu, W. Lu, L. Zhang, and N. Wang, K. Kuang, S. Wang, S. Packages 0 No packages published.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window.
0コメント