1. S MARTPHONES IMPACT IN 3G AND F UTURE LTE N ETWORKS Student : Adnan Basir (84906S) Supervisor : Jyri Hämäläinen Instructor : Timo Halonen (Nokia Siemens Networks ) Work carried out at Nokia Siemens Networks, Espoo Aalto University, School of Electrical Engineering 17-March-2011

2. Table of Contents  Introduction /Background  Reason for Smartphone Signaling Congestion  Problem statement  Layer 3 Radio Resource Control (RRC) Cell states.  Method Used  Important Results  Signaling Congestion in LTE  Conclusion and Future Work

3. Introduction  What is smartphone ?  A device that has ability to browse internet, send/receive email, edit office tasks etc, in addition to make regular phone calls.  What is Smartphone Signaling Challenge ?  Huge Influx of smartphones to the market  Congestion in cellular networks Failed call attempts High congestion report in newyork (AT&T case) and london  Is increase in data usage the real problem ? High data usage is only among small group of subscribers but congestion is in entire network.

4. Why Congestion then !  Signaling Traffic is the culprit.  30 – 50 % higher now as compared to data traffic.  Applications in smartphones connect to internet even when users are no using them (i.e. facebook updates, tweets etc).  Keep-alive messages  3G networks were designed to handle huge data traffic but not signaling traffic !  With influx of tablet devices the problem is going to get worse

5. RRC Cell states  What is signaling message ?  Each time a mobile make state transition it generates signaling messages.  30 messages to go between idle and dch  7 messages between dch and pch  2 messages between pch and fach  Some mobile vendors implement Fast dormancy which is transition from DCH to IDLE state when there is no data.

6. Thesis objective  Observe impact of various smartphones applications on network in terms of resource management.  Best network parameters for optimized services  T1, T2, T3 timers etc  To study How LTE cope with signaling congestion problem.

7. Method Used  3G Test Network  3G RRC logs with EMIL Tool (NSN propriety)  Wireshark for traces from 3G Gn Interface  NSN NETACT to configure network parameters  Nemo Handy (Used in case of outdoor map studies)

8. Important Results  Online Gaming  30 minutes game play  A single voice call would result in estimated 50 signaling messages.  Android uses “Fast dormancy” which results in high number of signaling messages G Android IPhone 4.1 345 1996 97 3201 TOTAL STATE TRANSITIONS TOTAL SIGNALING MESSAGES (Respective to state transitions)

9. Important Results RABS DCH>FACHFACH>PCHPCH>FACH FACH>DCH 97 3201 STATE TRANSITIONS SIGNALING MESSAGES (Respective to state transitions) SIGNALING IMPACT (TEXAS HOLDEM ON GOOGLE ANDROID) Total Signaling Messages = 3201 ! Android uses fast dormancy and forces the UE to idle state when there is no data ! 30 minutes game play ! RABS DCH>FACHFACH>PCHPCH>FACH FACH>DCH 2 66 156 936 18 36 14 28 155 930 STATE TRANSITIONS SIGNALING MESSAGES (Respective to state transitions) SIGNALING IMPACT (LIVE POKER ON IPHONE 4.1) Total Signaling Messages = 1996 !

10. 24 Hour user profiling with N97 In 24 Hour resources utilization estimate, idle, shoulder and busy cases have been divided as following User remain idle for 14 hours in 24 hours User perform shoulder activity for 7 hours in 24 hours. User perform busy activity for 3 hours in 24 hours For N97 mini, the results from 24 hour profile is given in the table Power Consumpti on Signaling Messages RABsDownlink Data Uplink Data TCP Sessions 1475 mA2486650449248 kbytes 3336 kbytes 902

11. Signaling Congestion in LTE  Simplified RRC Architecture  RRC Idle Cell selection Monitoring  RRC Connected DRX in connected state  Only one reconfiguration message  No RNC  Simple RRC connection procedure.  No NBAP and ALCAP signaling  Only PS Domain  No signaling connection release procedure  MME Handling most of the load now  So there can more signaling congestion on core network side

12. Other important studies  Impact of CELL_PCH  Laptop (with utms dongle) Vs Mobile signaling Load  Maps Studies (Google maps, Ovi maps)  Angry bird (Online Game)  MiFi study

13. Conclusion / Future Work  Implement HSPA Release 7 Cell-Fach  HS-DSCH instead of FACH  High efficiency  Implement Cell-Pch  Avoid fast-dormancy  Femtocells  Load sharing (signaling/data)  Intelligent network monitoring  Application using more signaling resources  Charge users on application bases

14. QUESTIONS ? THANK YOU !