Performance of interoperator fixed-mobile network sharing Ireneusz - PowerPoint PPT Presentation

Performance of interoperator fixed-mobile network sharing Ireneusz Szcześniak Institute of Computer and Information Sciences Częstochowa University of Technology, Poland Andrzej R. Pach Department of Communications AGH University of Science and Technology, Poland Bożena Woźna-Szcześniak Institute of Mathematics and Computer Science Jan Długosz University, Poland ONDM 2017 Grant number DEC-2013/08/S/ST7/00576 from the Polish National Science Centre

Introduction Contribution Evaluation Results Conclusions Introduction • Fixed-mobile networks are wide-spread, and expensive. • Operators cooperate and share to cut costs. • Currently sharing is limited to buildings, masts, etc. • We proposed the interoperator fixed-mobile network sharing and showed the splendid availability improvement. • We concentrate on passive optical networks (PONs). slide 1

Introduction Contribution Evaluation Results Conclusions Motivation • Performance improvement is crucial, especially for the next-generation wireless networks. • Performance improvement should be economical and scalable. • Interoperator fixed-mobile network sharing can improve performance by using redundant resources already deployed. slide 2

Introduction Contribution Evaluation Results Conclusions Redundancy is already there! Two separate PON deployments in the same area. slide 3

Introduction Contribution Evaluation Results Conclusions Contribution The performance evaluation of the interoperator fixed-mobile network sharing. The hallmark of our proposed sharing is the interoperator communication in access networks. slide 4

Introduction Contribution Evaluation Results Conclusions Interoperator fixed-mobile network sharing in general Internet IP network O1 O2 × default default × router router interoperator trunk aggregation Ethernet × × network switch access O1 O2 IC network FMN FMN slide 5

Introduction Contribution Evaluation Results Conclusions Interoperator sharing in passive optical networks IC IC CO CO IC slide 6

Introduction Contribution Evaluation Results Conclusions We need active nodes • In the proposed sharing we need active remote nodes. • Active nodes can diverge traffic to the other operator. • Active nodes are already used to, e.g., extend reach. slide 7

Introduction Contribution Evaluation Results Conclusions PON performance • We study the improvement of the PON performance. • Specifically, we study the upper bound of the improvement. • PON performance is the average ONU performance, and: load serviced by PONs ONU performance = load requested by the ONU • PON performance is a function of: • l - network load, • q - probability that a remote node is active, • r - probability that an ONU is interoperator-communicating. • We study two scenarios: • in the first, the locations of active remote nodes are given, • in the second, the active nodes are randomly distributed. slide 8

Introduction Contribution Evaluation Results Conclusions First scenario, and second too 1 − s 1 − s CO s s 1:g 1:g 1:g 1 − s 1 − s 1 st stage 2 nd stage 3 rd stage slide 9

Introduction Contribution Evaluation Results Conclusions Performance evaluation • Numerical evaluation: a mix of analysis and simulation. • We analytically evaluate a given, concrete network, for a given network load l . • Network load is evenly distributed among ONUs. • ONUs with less connectivity options are served first. • We randomly produce a sample of concrete networks from the populations with the given probabilities: • q - a remote node is active, • r - an ONU is capable of inter-operator communication. • We used 204600 concrete networks, and averaged the results. slide 10

Introduction Contribution Evaluation Results Conclusions The problem of shortest paths in PONs with active nodes IC-ONU NIC-ONU RN1 RN2 RN3 Example of finding a correct shortest path between an IC-ONU (interoperator-communicating ONU) and an NIC-ONU (non-iteroperator-communicating ONU), where • is a passive remote node, ◦ is an active remote node, is the correct path, and is the wrong path. slide 11

Introduction Contribution Evaluation Results Conclusions Graph model for shortest paths in PONs with active nodes IC-ONU NIC-ONU RN1 RN2 RN3 Example of modifying a graph by splitting passive remote nodes, where is a split passive remote node, ◦ is an active remote node, and is the correct path. slide 12

Introduction Contribution Evaluation Results Conclusions PON performance in the 1 st evaluation scenario 1 performance 0 . 9 0 . 8 0 . 7 0 . 6 0 . 5 2 1 . 5 10 0 10 − 1 1 10 − 2 l 10 − 3 r l - network load r - probability an ONU is capable of interoperator communication slide 13

Introduction Contribution Evaluation Results Conclusions PON performance in the 2 nd evaluation scenario for l = 2 1 performance 0 . 9 0 . 8 0 . 7 0 . 6 0 . 5 1 0 . 5 10 0 10 − 1 0 10 − 2 q 10 − 3 r q - probability a remote node is active r - probability an ONU is capable of interoperator communication slide 14

Introduction Contribution Evaluation Results Conclusions Conclusions • We evaluated the performance of the proposed sharing. • The performance could be improved twofold, or even more when there are more sharing operators. • Upgrades can be rolled out in stages and where needed most. • The proposed sharing significantly improves availability too. • There are many problems to research further, for instance: • implementation details, • sharing rules, • optimization. slide 15