PSTNOverviewPSTN Economic Regulation (US)Level Skipping vs. Dynamic Non-Hierarchical Routing (DNHR) 1980sDNHR – Flattening the Hierarchy (1980s)Call ScenariosArchitecture of Nano Network(in-network local calling)Architecture of Mini Network(in-network local calling)Nano vs. Mini NetworksWhat happens to Pearson’s if we had more routing information for Mini Bell?What happens to Pearson’s if we had more routing information for Mini Bell?Nano Connected to Mini Network(inter-network local or long distance)Robustness in Fiber Ringscollapsed vs. separate SONET ringsRobustness to Loss of Nodes or Edges in Nano BellRobustness to Loss of Nodes or Edges in Nano BellSummary of ConstraintsContributionsRecommendations for Future Work© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology1PSTNByJijun LinDan LivengoodChintan VaishnavMay 11th2006Faculty Advisor: Dan WhitneyESD 342, Advanced Systems Architecture© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology2Overview• A new historical perspective• Call scenarios and architectural comparison– Static: Network Metrics– Dynamic: Pearson’s Coefficient in evolving network• Robustness in new networks• Constraints and responses• Contributions and future work© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology3PSTN Economic Regulation (US)1934-1984(national)1984-1996(each region)After 1996(each state)AT&TAT&TMCISprintRegionalBells(Nine Total One perRegion)AT&TMCISprintILECCLECCLECCLECOur analysis focuses on an ILEC and a CLEC for a single stateMaxi BellMini BellNano BellFigure by MIT OCW.International gateway exchange (Centre de Transit 3)National tandem exchanges(Tertiary trunk switching centers)Regional tandem exchanges(Secondary trunk switching centers)Local tandem exchanges(Primary trunk switching centers)Local exchangesSubscriber linesSatellite linksSubmarinecablesInternationalnetworkTrunk networkLocalnetwork© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology4Level Skipping vs. Dynamic Non-Hierarchical Routing (DNHR) 1980sBefore DNHR -Level SkippingDNHRImproved quality via level skipping• Switch quality no longer main constraint • 5-level hierarchy structure no longer required • New constraints are the capacity on the links and switch reliability• Statistical analysis allows for dynamic planning of routes in pre-set time periodsClass 1Regional CenterClass 2Sectional CenterClass 3Primary CenterClass 4Toll CenterClass 5End OfficeFive-level toll switching plan in use from the 1950s.A variety of routings was possible with a maximumof nine trunks in tandem.Figure by MIT OCW. After Andrews & Hatch, 1971.© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology5DNHR – Flattening the Hierarchy (1980s)Architecture influenced by DNHR:Now Nano, Mini and Maxi can have similar architecture, but different coverageEnd System Backbone Central OfficesFigure by MIT OCW.REGIONAL CENTERSFigure by MIT OCW.© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology6Call ScenariosIn-networkLocal CallsInter-networkLocal & Long-distance CallsInter-networkLocal &Long-distanceCallsIn-networkLocal CallsNano MiniFive Networks2005 Nano Network2010 Nano NetworkMini Network2005 Nano + Mini2010 Nano + MiniNanoMiniBetween Nano, Mini and Nano + Mini networks, we can study all callscenarios© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology7Architecture of Nano Network(in-network local calling)Nano Bell’s Plan for Migration from 2005 to 20101. Get redundant fiber outlets2. Get every node on fiber (preferably ring)N = 104 z = 2.327M = 121 l = 7.308C = 0.0262N = 123 z = 2.452M = 152 l = 8.729C = 0.020620052010© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology8Architecture of Mini Network(in-network local calling)N = 171 z = 5.218M = 446 l = 2.582C = 0.1179Four Types of TandemsLocalAccessToll911Why are Nano and Mini different?Legacy ArchitectureRegulatory ObligationsVoice vs. Data NetworkIs there a parameter that indicates the difference in Nano vs. Mini Network?Nano vs. Mini NetworksParameter Nano 2005 Nano 2010 Mini OnlyN 104 123 171M 121 152 446z (<k>) 2.327 2.452 5.216l 7.308 8.729 2.582log n/ log <k> 5.499 5.365 3.113C 0.0262 0.0206 0.1179<k>/n 0.022 0.020 0.031r 0.2196 0.3277 -0.6458Is sharply different rindicative of differencesin technology?Perhaps not….We know from level-skipping and DNHR that Central Offices (not just the tandems) are connected in Mini’s network, so r must be higher. But we simply can’t get this information because of privacy/competitive reasons.© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology10What happens to Pearson’s if we had more routing information for Mini Bell? r = -0.6458r = 0.7403© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology11What happens to Pearson’s if we had more routing information for Mini Bell?• Degree correlation changes from -0.6458 to 0.7403 by randomly adding 0 up to 1755 edges)• On average, zero degree correlation happens at adding 185 edges0 500 1000 1500 2000-0.8-0.6-0.4-0.200.20.40.60.8Number of randomly added linksPearson degree correlationSimulation of pearson degree correlation(average over 50 runs)add 182 edges (10.4%)© 2006 Student: Jijun Lin, Daniel Livengood, Chintan Vaishnav, Engineering Systems Division, Massachusetts Institute of Technology12Nano Connected to Mini Network(inter-network local or long distance)20052010Parameter Nano 2005 Nano 2010 Mini Only Mini+Nano 2005 Mini+Nano 2010N 104 123171 275 295M 121 152446 667 714z (<k>) 2.327 2.4525.216 4.85 4.84l 7.308 8.7292.582 3.71 4.275log n/ log <k> 5.499 5.365 3.113 3.557 3.606C 0.0262 0.02060.1179 0.196 0.2136<k>/n 0.022 0.020 0.031 0.018 0.016r 0.2196 0.3277-0.6458 -0.1882 -0.1552All critical measures of Nano + Mini fall in between