Mobile Network Sharing Mobile Network Sharing Between Operators: Between Operators A Demand Trace-Driven Study Di Francesco et al. Outline Paolo Di Francesco, Francesco Malandrino, Luiz DaSilva Introduction Data Available Data Trinity College Dublin Coverage Correlation Time Space Assessing the effectiveness Results Conclusion & Future Works Monday 18 th August, 2014 the telecommunications research centre
Mobile Network Outline Sharing Between Operators Di Francesco et al. Outline Introduction Data Available • Introduction. Data Coverage • Data. Correlation Time • Correlation (time and space). Space • Effectiveness of network sharing. Assessing the effectiveness Results • Conclusion & future works. Conclusion & Future Works
Mobile Network Long story short Sharing Between Operators Di Francesco et al. Outline Introduction • Network sharing to reduce Data Available CAPEX and OPEX. Data + + Coverage MNO 1 MNO 1 • When does resource Correlation Time sharing make sense? Space Assessing the • How to assess the MNO 2 = MNO 2 = effectiveness effectiveness of network Results Conclusion & sharing? Future Works MNO 1 +MNO 2 MNO 1 +MNO 2
Mobile Network Data Available Sharing Between Operators Data made available by two operators: Meteor & O2 Di Francesco et al. • Transmitters’ location spread throughout the Republic of Outline Ireland ( > 20.000). Introduction • Coverage (approx.) & sectorization information. Data Available Data Coverage • Detailed call records & data sessions (many million). Correlation • Long observation time (weeks). Time Space • Information on mobile device types usage Assessing the effectiveness (TAC codes > 3.000). Results Conclusion & Future Works Limitations: • Partial mobility information. • Users precise location is unknown. • Datasets have been collected at different time frame.
Mobile Network GSM coverage Sharing Between Operators Di Francesco et al. MNO 1 MNO 2 Outline Introduction Data Available Data Coverage Correlation Time Space Assessing the effectiveness Results Conclusion & Future Works
Mobile Network 3G coverage Sharing Between Operators Di Francesco et al. MNO 1 MNO 2 Outline Introduction Data Available Data Coverage Correlation Time Space Assessing the effectiveness Results Conclusion & Future Works
Mobile Network Autocorrelation Sharing Between Operators Di Francesco et al. Outline Voice traffic Introduction Data Available 1 MNO 1 − Aggregated Data MNO 2 − Aggregated 0.8 Coverage MNO 1 − Busiest sector 0.6 MNO 2 − Busiest sector Correlation MNO 1 − Median sector 0.4 Time Autocorrelation MNO 2 − Median sector Space 0.2 Assessing the 0 effectiveness −0.2 Results −0.4 Conclusion & −0.6 Future Works −0.8 −1 0 20 40 60 80 100 120 140 160 Hours [h]
Mobile Network Autocorrelation Sharing Between Operators Di Francesco et al. Outline Voice traffic Data traffic Introduction Data Available 1 1 MNO 1 − Aggregated MNO 1 − Aggregated Data MNO 2 − Aggregated MNO 2 − Aggregated 0.8 0.8 Coverage MNO 1 − Busiest sector MNO 1 − Busiest sector 0.6 0.6 MNO 2 − Busiest sector MNO 2 − Busiest sector Correlation MNO 1 − Median sector MNO 1 − Median sector 0.4 0.4 Time Autocorrelation MNO 2 − Median sector Autocorrelation MNO 2 − Median sector Space 0.2 0.2 Assessing the 0 0 effectiveness −0.2 −0.2 Results −0.4 −0.4 Conclusion & −0.6 −0.6 Future Works −0.8 −0.8 −1 −1 0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160 Hours [h] Hours [h]
Mobile Network Space correlation (1) Sharing Between Operators Di Francesco 1 . 0 et al. MNO 1 - weekend MNO 2 - weekend Global (Moran’s Index): Spatial correlation [Global Moran’s Index] MNO 1 - weekday Outline MNO 2 - weekday 0 . 5 Introduction 0 . 0 Data Available Data Coverage − 0 . 5 n n Correlation w i , j ( x i − ¯ X )( x j − ¯ � � X ) Time i =1 j =1 , j � = i n Space − 1 . 0 n p S 0 6am 12am 6pm 12am ( x i − ¯ X ) 2 Time of day � Assessing the effectiveness i =1 1 . 0 MNO 1 - weekend Results MNO 2 - weekend Spatial correlation [Global Moran’s Index] MNO 1 - weekday Conclusion & w i , j = | A i ∩ A j | MNO 2 - weekday 0 . 5 Future Works | A i ∪ A j | 0 . 0 − 0 . 5 − 1 . 0 p 6am 12am 6pm 12am Time of day
Mobile Network Space correlation (2) Sharing Between Operators Di Francesco 120 MNO 1 , w/o sharing et al. MNO 2 , w/o sharing Local: 100 MNO 1 , w/ sharing MNO 2 , w/ sharing Outline It allows to label 80 Number of hot spot Introduction transmitters as: outlier 60 Data Available (hot, cold), cluster (hot, Data 40 cold) Coverage 20 Correlation Time n 0 x i − ¯ Space w i , j ( x j − ¯ X 6am 12pm 6pm 12am � X ) Time of day S 2 Assessing the i j =1 , j � = i effectiveness Results 60 MNO 1 , w/o sharing w i , j = | A i ∩ A j | MNO 2 , w/o sharing Conclusion & 50 MNO 1 , w/ sharing Future Works | A i ∪ A j | MNO 2 , w/ sharing 40 Number of hot spot 30 20 10 0 6am 12pm 6pm 12am Time of day
Mobile Network Hot Spots in Dublin - 24h Sharing Between Operators timelapse Di Francesco et al. Outline Introduction Data Available Data Coverage Correlation Time Space Assessing the effectiveness Results Conclusion & Future Works
Mobile Network Wrapping up the results Sharing Between Operators Di Francesco et al. Outline Introduction Data Available Operator Ireland Urban Rural Data Coverage Deployment density MNO 1 0.080 4.488 0.040 [sectors/km 2 ] MNO 2 0.095 5.615 0.042 Correlation MNO 1 0.10 0.08 0.11 WE Time Space correlation MNO 2 0.13 0.11 0.25 Space [Moran’s Index] MNO 1 0.07 0.08 0.10 WD MNO 2 0.04 0.04 0.11 Assessing the effectiveness MNO 1 -55% -38% -93% WE hot spot MNO 2 -55% -35% -50% Results reduction MNO 1 -64% -46% -96% WD Conclusion & MNO 2 -54% -44% -93% Future Works
Mobile Network Future works Sharing Between Operators Di Francesco et al. Outline Introduction Data Available Data • Combine demand data with census data publicly available. Coverage Correlation • Propose actual mechanisms for effecting network sharing. Time Space • Energy efficiency, cost reduction, through network sharing. Assessing the effectiveness Results Conclusion & Future Works
Mobile Network Sharing Between Operators Di Francesco et al. Outline Introduction Thank You Data Available Data Coverage Correlation Time Space Assessing the effectiveness Results Conclusion & Future Works
Recommend
More recommend