Towards Eco-Friendly Home Networking Mathias Gibbens, Chris Gniady - PowerPoint PPT Presentation

Towards Eco-Friendly Home Networking Mathias Gibbens, Chris Gniady and Beichuan Zhang Department of Computer Science The University of Arizona Tucson, Arizona IGCC '14 1

Home networks are complex NAS ● More and more demand is being placed on router performance ● Routers require more computing power, bandwidth and features IGCC '14 2

Complexity → Power Power WiFi Speed CPU Dual core! RAM g (2003) n (2009) ac (2013)  Power consumption doubled in 5 years, what about the future?  Always on in 88 million homes, energy footprint of $1 billion IGCC '14 3

Previous solutions ● Wired: IEEE 802.3az introduced Energy-Efficient Ethernet ▪ First deployed in home networks ▪ Physical connection, easy to detect client ● Large mesh networks: many routers, many clients ▪ Power down redundant access points ▪ Client picks optimal network when more than one is available ● Home networks: one router, many clients ▪ Goma et al: aggregate individual networks ▪ Requires: dense networks, cooperation, client modifications We need energy management for individual routers IGCC '14 4

Our contribution ● Transparent energy optimization of personal networks ▪ Individual routers ▪ No user intervention or modification of clients ▪ No cooperation between networks ● Implementation approach ▪ Discarding unnecessary wireless traffic ▪ Powering down routers when idle ▪ Power cycling with active clients ● Increased energy efficiency of individual home routers IGCC '14 5

Outline  Introduction  Trace collection and categorization  Proposed optimizations  Methodology  Results  Conclusion IGCC '14 6

Traffic categorization No client broadcast No client Idle client Active client 100% 80% 60% Time 40% 20% 0% T1 T2 T3 T4 T5  Traffic seen even without clients connected  Lots of idle time when clients are present  Router is in full power mode independent of clients IGCC '14 7

Eliminating broadcast traffic 100000 10000 1000 100 10 1 Client Broadcast  Broadcast traffic, but no clients around to respond  Traffic from wired interface retransmitted over wireless  No one listening → drop broadcast traffic  Safe to perform: clients must be present in order to respond IGCC '14 8

Powering down when idle 100000 10000 1000 100 10 1 Client Broadcast Last client departs Power down Look for clients Power Power up  No clients → power down antennas after a timeout  Periodically check for the arrival of new clients IGCC '14 9

Optimizing duty cycle: Downtime 30 Connection time [s] 20 10 0 0 1 2 3 4 5 6 7 8 9 10 11 Antenna down time [s] ● Duty cycle impacts a client's ability to connect ● Need to balance extra delay with potential energy savings ● Infrequent initial associations, which impact only first client IGCC '14 10

Optimizing duty cycle: Uptime 30 Connection time [s] 20 10 0 4 5 6 7 8 9 10 11 Antenna up time [s] ● Antennas must be up for at least 4 seconds for clients to connect ● From observed delays, we chose a 5/5 second up/down cycle IGCC '14 11

Idle connected clients 100000 10000 1000 100 10 1 Client Broadcast ● During sufficiently long idle periods, turn off transmit antenna ● Possible because either clients initiate or data arrives on wire ● Router must still periodically announce its presence IGCC '14 12

Idle connected clients ACK Time No ACK, try resending Packet lost, inform user ● WiFi spectrum is inherently error prone ▪ Existing protocols have built in transparent retransmission to compensate ● Clients typically disconnect from network after 7 seconds ● Transparent retransmission gives us opportunity to power up IGCC '14 13

Idle connected clients Response to client Complete Client packet Begin router power up Response Router power Forward packet ● Additional delay due to power up may be seen by applications ▪ Can be hidden in the time it takes for a response to return to the router ● Transitioning router's state has an energy cost ▪ Only sufficiently long periods should be optimized ▪ Ensures real time data is not interrupted IGCC '14 14

Methodology Trace T1 T2 T3 T4 T5 Average concurrent devices 1 5 4 4 2 Maximum concurrent devices 1 9 7 6 3 Initial associations 13 16 14 31 35 Average time with no client [h] 10.7 0.08 2.07 1.32 0.92 Traffic volume [GB] 5.57 40.21 4.22 3.52 12.97  Monitor traffic seen at router: week long traces  Very few initial associations when no other clients present  Detailed power/delay model of ASUS RT-N16 profiled using NI IGCC '14 15

Eliminating broadcast traffic No client broadcast No client Idle client Active client 100% 80% 60% Time 40% 20% 0% T1 T2 T3 T4 T5  No client period increased by 10% average in traces 2-5  More opportunities for router to power down IGCC '14 16

Power cycling no clients No client Idle client Active client  Power cycling with no clients 0.5 has significant impact 0.4  Additional connection delay for first client paid only 0.3 ] occasionally J M [ y g 0.2 r e n E 0.1 0.0 NB PC NB PC NB PC NB PC NB PC T1 T2 T3 T4 T5 NB – No Broadcast PC - PowerCycle IGCC '14 17

Active client optimizations No client Idle client Active client  More aggressive power 0.5 cycling can reduce energy 0.4 consumption by an additional 20-30% 0.3 ]  Cumulative energy savings J M [ y observed to be 12-59% g 0.2 r e n E 0.1 0 C T A C T A C T A C T A C T A C C C C C C C C C C P P P P P T1 T2 T3 T4 T5 PC – PowerCycle CT – CycleTransmit CA - CycleAll IGCC '14 18

Delay due to state transition  Active clients see some T1 T2 T3 T4 T5 100% delay when initiating activity after a period of 80% idle time  All delays within n 60% e e s perception threshold, not s y 40% a l noticed by user e D 20% 0% 0 10 20 30 40 50 60 70 80 Additional delay [ms] IGCC '14 19

Conclusion  Investigated opportunities to reduce energy consumption of consumer wireless routers  Collected traces from personal networks  Predicted wireless energy consumption reduced by 12-59%  Changes do not break backwards compatibility IGCC '14 20

Thank You Questions? IGCC '14 21

Trace collection and analysis 10000 1000 100 10 1  Five unique week-long traces collected from households  Routers recorded just the wireless traffic seen  Networks used as normal to produce representative traces IGCC '14 22

Idle periods with clients  To save energy with clients, T1 T2 T3 T4 T5 100% there must be many idle periods of sufficient length 80% e  Each trace has many long m 60% i t e idle periods l d i 40% d e t h g 20% i e W 0% 4 16 64 256 1024 2 8 32 128 512 >1024 -20% Idle time [s] IGCC '14 23