1. Completed/planned activities of AGH in VD-A Miroslaw Kantor, Marcin Niemiec Department of Telecommunications AGH University of Science and Technology VD-A meeting February 26, 2007 Barcelona, Spain

2. February 26, 2007 Barcelona, Spain Completed/planned activities Access network resilience (will be continued) Network protection strategies for access networks Analytical reliability analysis of access networks Stochastic simulation for evaluation of access networks reliability/availability System reliability design problem Quantum cryptography (planned) Study of optical network technologies for adaption of quantum cryptography techniques to optical access networks

3. February 26, 2007 Barcelona, Spain Stochastic simulation approach n In most of the practical cases we cannot specify precisely the model or the value of its parameters n The assumption of deterministic values for the uncertainty might lead to faulty models whose analysis might not be in conformity with reality n Parameter uncertainty is typically specified in terms of probability density functions PDFs

4. February 26, 2007 Barcelona, Spain Considered types of system performance: s-expected system lifetime: mean system lifetime (expected mean value EMV)  -system lifetime: denotes the largest satisfying the inequality: This means that reaches upwards of with a probability  system reliability: probability that the system lifetime is larger than or equal to the given value Types of system resilience performance

5. February 26, 2007 Barcelona, Spain Passive Optical Network recovery schemes [G.981.5] Recovery architecture of type A Recovery architecture of type B Recovery architecture of type CRecovery architecture of type D

6. February 26, 2007 Barcelona, Spain Reliability block diagrams for PON recovery schemes RBD for recovery architecture of type A RBD for recovery architecture of type B RBD for recovery architecture of type C RBD for recovery architecture of type D: duplex RBD for recovery architecture of type D: no duplex

7. February 26, 2007 Barcelona, Spain Assumptions for stochastic simulation No element repair action is taken after the failure The exponential distribution, as one of the most common distributions used in the systems reliability analysis, has been applied The length of the connection link is 20 km n Survivability parameters of the optical access network components: ElementFIT MTTF [year] OLT equipment20000 5  10 4 Optical fiber500 [FIT/km] 2  10 6 [h  km] Optical splitter120 8.3  10 6 ONU equipment8500 1.2  10 5

8. February 26, 2007 Barcelona, Spain Stochastic simulation results (basic architecture) PDF of system lifetime Stochastic simulation process for EMV

9. February 26, 2007 Barcelona, Spain Results for considered recovery architectures Considered architecture expected system lifetime EMV system reliability [T 0 =EMV] basic architecture26029119110.3662 type A architecture32234156670.3716 type B architecture45419292900.4018 type C architecture51278356320.4054 type D arch. duplex66391467400.4111 type D arch. no duplex45222299530.4062 M. Kantor, K. Wajda, P. Cholda: „Reliability Evaluation of PON Recovery Architectures Using Stochastic Simulation”, BBEurope,11-14 December 2006,Geneva, Switzerland

10. February 26, 2007 Barcelona, Spain Planned activity Study of optical network technology and architectures (PON, WDM...) in order to adapt quantum cryptography techniques to optical access networks

11. February 26, 2007 Barcelona, Spain Quantum Cryptography Quantum Cryptography (QC) provides unbreakable communications – each eavesdroped intruders are detected. QC solves key distribution problem over optical fiber networks. QC ensures absolute security by using the laws of quantum physics (a bits are coding by quantum states of photons). Currently, QC devices are producing and testing (i.e. in Europe by Id Quantique or SmartQuantum) but these devices communicate upon a dedicated fibre. One of possible applications: FTTH PON technology uses broadcast downlink communication - each entity receives all the network traffic. Therefore, encryption in these networks is crucial in order to avoid frauds.

12. February 26, 2007 Barcelona, Spain Possible aspects Possible aspects to consider: Analysis of necessity Analysis of technical possibility Requirements Security issues Services Business issues Case studies … Contact person: Marcin Niemiec niemiec@kt.agh.edu.pl

13. February 26, 2007 Barcelona, Spain Planned joint activity Topic : Experimental evaluation of access network resilience schemes Paticipants: FUB, AGH Planned mobility: October 2007

14. Thank you!

15. February 26, 2007 Barcelona, Spain Conclusions n Different access architectures based on the Passive Optical Network concept have been compared n The comparison was made taking into account three types of system performance measures: s-expected system lifetime,  -system lifetime and system reliability n The stochastic simulation was applied for simple PON architecture to check the applicability of this method for reliability analysis of more complicated systems n The obtained results allow us to state that stochastic simulation method can be used for that purpose

16. February 26, 2007 Barcelona, Spain System lifetime - random lifetime of elements j working in components i For a parallel redundant system the lifetimes of components: For a standby redundant system the lifetimes of components:

17. February 26, 2007 Barcelona, Spain Stochastic simulation for EMV A structure of considered protection architecture has to be decided Vectors of the lifetimes, from the lifetimes of redundant element have to be generated n The bisection search can be used to estimate n The mean system lifetime:

18. February 26, 2007 Barcelona, Spain Stochastic simulation for  -system lifetime N s-independent random vectors are generated n Bisection search is used to approximate for each n After simple data processing, a sequence of lifetimes ordered from the smallest to the biggest is achieved n Let the integer part of to be denoted by n Considering the law of large numbers it can be stated, that the N’th largest element in the obtained sequence is the estimation of  -system lifetime,

19. February 26, 2007 Barcelona, Spain Stochastic simulation for system reliability N s-independent random vectors are generated n Bisection search is used to approximate for each, n After simple data processing, a sequence of lifetimes ordered from the smallest to the biggest is achieved n The number of cases when following condition occurs:, is denoted by N’ n Considering the law of large numbers it can be stated, that the system reliability can be estimated by the ratio of :