MIT CSAIL 1 Problem: 3G/LTE is a battery hog Up to 14 hours on 2G - PowerPoint PPT Presentation

Shuo Deng, Hari Balakrishnan MIT CSAIL 1

Problem: 3G/LTE is a battery hog  “Up to 14 hours on 2G” “Up to 6.5 hours on 3G” 100% 90% 80% 70% Energy Percentage 60% 50% 40% 30% 20% 10% 0% Goal: Reduce Energy News IM Micro-blog Game Email Social Finance Network Consumption Data Active, No Data High PowerIdle State Switch 2

Context: Radio Resource Control Inactivity Timer t 1 Cell_DCH (Active) Cell_FACH Cell_PCH/IDLE (High Power Idle) (Idle) Inactivity Timer t 2 3

t 1 Active Power Consumption High Power IDLE Idle t 2 t 1 t 2 IDLE IDLE 4

Fast Dormancy Inactivity Timer t 1 Active High Power Idle Idle Inactivity Timer t 2 Idle Active 5

Idle Active Challenges  Switching between states takes time(1~3 seconds), and consumes energy  Signaling overhead t1 t2 IDLE IDLE 6

Contributions  A traffic-aware design to control radio state transitions to reduce energy consumption  MakeIdle: when to switch to Idle  MakeActive: when to switch to Active MakeActive Idle Active MakeIdle  Experimental evaluation on real usage data  Energy reduction up to 75% across different carriers 7

System Design Active Idle App1 App2 App3 Control Socket Module Calls Packet/Socket Call Switch to Idle Information Modified Socket Layer Switch to Active Fast Data Dormancy Interface Trigger Fast Dormancy 3G Radio 8

t1 t2 IDLE IDLE Power time 9

Active Idle MakeIdle Algorithm MakeIdle Power Power time time IAT Inter Arrival Time > , should switch to Idle mode to If minimize the energy consumption. If IAT > threshold , should switch to Idle mode. 10

Active Idle MakeIdle Algorithm MakeIdle  Predict whether the IAT will be greater than threshold  Wait for a short period of time t wait , if no packet comes, then put the radio to Idle mode  Why: the longer the network is idle, the longer it is likely to remain idle 11

Active Idle MakeIdle: Picking t wait MakeIdle … P(IAT>t+threshold | IAT>t ) Previous N packets   ( Energy Energy ) state _ switch no _ state _ switch 12

Power Current Scheme time Power With MakeIdle time Power With MakeActive time 13

MakeActive Active Idle MakeActive Algorithm  Reduce the number of state switches by introducing a small delay when the radio is in Idle mode and data transmission requests come from the mobile device side  How much delay for each request?  Fixed delay bound  Learning algorithm 14

Evaluation Setup  Energy profiling  Power consumption profiles for 4 US major carriers: AT&T, Verizon, T-Mobile, Sprint  Trace driven simulation  Tcpdump traces for real usage data, collected from 9 users, 28 days in total 15

Evaluation: MakeIdle 40x 80 95% IAT across users 35x 70 Energy Saved (%) Increased Signaling 95% IAT per user 30x 60 MakeIdle Overhead 25x 50 Prior knowledge of IAT 40 20x (Oracle) 30 15x 20 10x 10 5x 0 0 1 2 3 -10 1 2 3 User ID User ID 16

Active Idle Evaluation: MakeIdle MakeIdle  Prediction Accuracy 95% IAT across users FP 95% IAT across users FN 95% IAT per user FP 30 95% IAT per user FN False Positive (FP) or 25 False Negative (FN) MakeIdle FP MakeIdle FN 20 15 10 5 0 1 2 3 User ID 17

Evaluation: Different Carriers 80 70 Energy Saved (%) 60 95% IAT across users 50 95% IAT per user 40 MakeIdle 30 MakeIdle+MakeActive 20 10 0 T-Mobile AT&T Verizon 3G Verizon LTE -10 Carriers 18

Related Work  Inactivity timer reconfiguration  Statistical method [Falaki et al, 2010]: 95 percentile packet inter arrival time  Applications-Involved Design  TailEnder [Balasubramanian et al, 2009]: each application specifies its delay tolerance  TOP [Qian et al, 2010]: application predict the gap between its own traffic transmissions  TailTheft [Liu et al, 2011]: application specifies delay tolerance and predicts transmission duration 19

Conclusion  A traffic-aware design to control state transitions of 3G/LTE radio to reduce energy consumption on mobile devices  Require no modifications of the applications  Save 3G/LTE energy consumption by up to 75% across different carriers 20