Optical packet switches that resolve contention by recirculating packets in a WDM shared loop buffer are validated only at the single-node level: the maximum number of allowed circulations is set by accumulated amplifier noise and crosstalk, and packet-loss and delay results come from isolated-switch models. The open problem, posed by the architecture's originators and left standing by the follow-up literature, is the coupled physical-layer/network-layer analysis of a network built from such switches: how a packet's circulation history accumulated along a multi-hop route depletes its remaining allowable circulations and end-to-end bit error rate at downstream nodes, where and when optical regeneration should be placed and how it reshapes end-to-end error and buffering budgets, and how these effects jointly determine system-level packet loss, delay, and the required count of tunable wavelength converters and other optical components. A 2018–2026 continuation stream has covered first-order cascade BER, queueing-level network loss/delay, and single-node in-buffer regeneration, but no published work models history-dependent circulation budgets or network-wide regeneration placement, so the integrated network-scale question remains open.

Background

Optical packet switching (OPS) is a proposed transport technology in which data packets are routed entirely in the optical domain, avoiding optical-electrical-optical conversion; its central difficulty is contention, which arises when two packets simultaneously request the same output port. Because no practical optical random-access memory exists, OPS architectures buffer packets in fiber delay lines (FDLs) — fixed fiber lengths that delay a packet by a fixed time. In a recirculating (feedback) loop buffer, a contending packet is sent around a fiber loop and re-enters the switching fabric in a later slot; using wavelength division multiplexing (WDM, many wavelength channels in one fiber) inside the loop lets many packets share one physical loop on distinct wavelengths. Moving a packet between wavelengths requires a tunable wavelength converter (TWC), and passive wavelength-selective routing is typically provided by an arrayed waveguide grating (AWG), a device whose input-to-output routing map is fixed by the signal wavelength. Erbium-doped fiber amplifiers (EDFAs) compensate the losses of each traversal, but every EDFA adds amplified spontaneous emission (ASE) noise and every component contributes crosstalk, so the bit error rate (BER) degrades with each circulation; this sets a circulation limit, a maximum number of times a packet may traverse the loop before its signal quality is unacceptable. WDM packet switches of this lineage were analyzed from the mid-1990s (Danielsen et al. 1997 showed how TWC sharing governs switch performance) and were carried toward system scale in the European IST DAVID project (Dittmann et al. 2003). Within this lineage, Singh, Srivastava and Singh (2007) proposed a WDM loop-buffer switch in the feedback shared-buffer configuration: contending packets enter a WDM loop buffer module, and TWCs place them on wavelengths that route them to free output slots via the AWG-based core. The paper derives the physical-layer noise model that fixes the maximum allowed number of circulations, and validates packet-loss and delay behavior with a mathematical model and simulation — but only for an isolated single switch. The paper itself identifies what this leaves unanswered at network scale: in a multi-hop network a packet may recirculate at several successive nodes, so link attenuation and noise picked up at upstream nodes reduce the circulation budget still available downstream; signal degradation from accumulated ASE and device noise may force optical regeneration somewhere in the network, but neither the placement nor the end-to-end BER and buffering consequences of regeneration were modeled; the effect of cascading many such switches and their interconnecting links on per-packet circulation budgets and on system-wide packet loss and delay was not analyzed; and the practicality of the large TWC and component counts (cost, footprint, control complexity) remained open. These four items form one question: extending the validated single-node loop-buffer analysis to a network of such switches.

Problem Statement

For a multi-hop network whose nodes are the WDM loop-buffer optical packet switches of Singh, Srivastava & Singh (2007), determine the system-level behavior that the single-switch analysis cannot: (i) how a packet's circulation history — the recirculations it has already made at upstream nodes, together with attenuation and noise accumulated on the inter-node links — constrains the number of circulations still allowable at each downstream node and its end-to-end BER; (ii) where and when in such a network optical signal regeneration should be performed, and how regeneration placement changes end-to-end BER and the buffering (circulation) budgets available network-wide; (iii) how these coupled effects determine system-level packet loss and delay as functions of the number of cascaded switches, link properties, and traffic load; and (iv) what these results imply for the required numbers of tunable wavelength converters and other optical components (cost, footprint, control complexity). A complete answer is a coupled physical-layer/network-layer model — reducing to the published single-switch noise and loss model when the network has one node — together with the resulting analysis or design rules, treating all four parts jointly: (i) the per-node constraint that accumulated circulations and inter-node attenuation and noise place on the remaining allowable circulations and on end-to-end BER; (ii) where and when optical regeneration should be performed, and how its placement changes end-to-end BER and the network-wide circulation budgets; (iii) the resulting system-level packet loss and delay as functions of the number of cascaded switches, link properties, and traffic load; and (iv) the implications for the required numbers of tunable wavelength converters and other optical components (cost, footprint, control complexity). Parts (i)–(iii) are the required modelling results; part (iv) is a qualitative design implication to be drawn from them rather than a separately modelled quantity.

References

  1. Rajat Kumar Singh, Rajiv Srivastava, Yatindra Nath Singh (2007). Wavelength division multiplexed loop buffer memory based optical packet switch. Optical and Quantum Electronics 39(1):15–34. DOI 10.1007/s11082-007-9061-0, https://doi.org/10.1007/s11082-007-9061-0
  2. Rajat Kumar Singh, Rajiv Srivastava, Yatindra Nath Singh (2007). AWG and EDFA based optical packet switch using feedback shared loop buffer memory. Optical and Quantum Electronics 39(14):1153–1165. DOI 10.1007/S11082-007-9147-8, https://doi.org/10.1007/S11082-007-9147-8
  3. Rajat Kumar Singh, Rajiv Srivastava, Yatindra Nath Singh (2007). Optical Packet Switch Architectures: A Comparative Analysis for Bursty Traffic and Optical Cost. Fiber and Integrated Optics 26(6):321–334. DOI 10.1080/01468030701595597, https://doi.org/10.1080/01468030701595597
  4. Rajiv Srivastava, Rajat Kumar Singh, Yatindra Nath Singh (2008). Scalability analysis of optical packet switch architectures. TENCON 2008 – IEEE Region 10 Conference. DOI 10.1109/TENCON.2008.4766405, https://doi.org/10.1109/TENCON.2008.4766405
  5. Vaibhav Shukla, Ramesh Kumar Sonkar, Rajiv Srivastava (2019). Performance comparison between recirculating loop buffer-based optical packet switch architectures. Indian Journal of Physics 94(7):1085–1096. DOI 10.1007/s12648-019-01540-5, https://doi.org/10.1007/s12648-019-01540-5
  6. S. L. Danielsen, B. Mikkelsen, C. Joergensen, T. Durhuus, K. E. Stubkjaer (1997). WDM packet switch architectures and analysis of the influence of tunable wavelength converters on the performance. Journal of Lightwave Technology 15(2):219–227. DOI 10.1109/50.554327, https://doi.org/10.1109/50.554327
  7. L. Dittmann, C. Develder, D. Chiaroni, F. Neri, F. Callegati, W. Koerber, A. Stavdas, M. Renaud, A. Rafel, J. Sole-Pareta, W. Cerroni, N. Leligou, L. Dembeck, B. Mortensen, M. Pickavet, N. Le Sauze, M. Mahony, B. Berde, G. Eilenberger (2003). The European IST project DAVID: a viable approach toward optical packet switching. IEEE Journal on Selected Areas in Communications 21(7):1026–1040. DOI 10.1109/JSAC.2003.816388, https://doi.org/10.1109/JSAC.2003.816388
  8. Ravi Kumar, Anurag Tripathi (2018). Cross-Layer Optimization and Cascadability of Optical Switches in Fiber Optic Data Networks. Journal of Optical Communications 42(4):575–584. DOI 10.1515/JOC-2018-0142, https://doi.org/10.1515/JOC-2018-0142
  9. Anjaney Nigam, Bharat Mishra, Prabhat Patel (2018). Recirculating Buffer and Arrayed Waveguide Grating-Based Switch for Optical Data Centers. Journal of Optical Communications 42(2):217–223. DOI 10.1515/JOC-2018-0067, https://doi.org/10.1515/JOC-2018-0067
  10. Arunendra Singh, Amod Kumar Tiwari (2018). Analysis of Hybrid Buffer Based Optical Data Center Switch. Journal of Optical Communications 42(3):415–424. DOI 10.1515/JOC-2018-0121, https://doi.org/10.1515/JOC-2018-0121
  11. Arunendra Singh, Amod Kumar Tiwari, Pronaya Bhattacharya (2019). Bit Error Rate Analysis of Hybrid Buffer-Based Switch for Optical Data Centers. Journal of Optical Communications 43(3):339–346. DOI 10.1515/JOC-2019-0008, https://doi.org/10.1515/JOC-2019-0008
  12. Vaibhav Shukla, Prabhat Kumar Patel, Dilip Kumar Choubey, Santosh Kumar (2020). Cascadability analysis of WDM recirculating loop buffer-based switch in optical data networks. Journal of Optical Communications 44(s1):s945–s953. DOI 10.1515/joc-2020-0100, https://doi.org/10.1515/joc-2020-0100
  13. A. V. Polyakov (2020). Simulation of Fiber-Optic Buffer Loop Memory with All-Optical 2R Regeneration. Optical Memory and Neural Networks 29(2):100–109. DOI 10.3103/S1060992X20020101, https://doi.org/10.3103/S1060992X20020101
  14. Arunendra Singh, Amod Kumar Tiwari, Vinay Kumar Pathak, Pronaya Bhattacharya (2021). Blocking performance of optically switched data networks. Journal of Optical Communications 44(s1):s1029–s1034. DOI 10.1515/JOC-2020-0263, https://doi.org/10.1515/JOC-2020-0263
  15. Preeti Singh, J. K. Rai, Ajay K. Sharma (2022). Packet blocking performance of optical networks under contention resolution mechanisms. Journal of Optics 52(3):1206–1217. DOI 10.1007/s12596-022-00938-5, https://doi.org/10.1007/s12596-022-00938-5
  16. Amit Sinha, Diwakar Bhardwaj, Vaibhav Shukla (2026). Performance analysis of optical communication networks with cascaded buffered optical switches. Journal of Optical Communications. DOI 10.1515/joc-2026-0218, https://doi.org/10.1515/joc-2026-0218
  17. M. A. Mestre, P. Jennevé, H. Mardoyan, A. Ghazisaeidi, S. Bigo, G. de Valicourt (2015). On the SOA cascadability and design rules for optical packet-switched networks. Optical Fiber Communication Conference (OFC) 2015, paper Th2A.2. DOI 10.1364/ofc.2015.th2a.2, https://doi.org/10.1364/ofc.2015.th2a.2
  18. Xuan-Chao Huang, Jay Cheng (2010). Constructions of Optical Queues With a Limited Number of Recirculations – Part II: Optimal Constructions. arXiv:1004.4277, https://arxiv.org/abs/1004.4277