Distribute what you can, centralize what you must

Download Report

Transcript Distribute what you can, centralize what you must

Distribute what you can,
centralize what you must!
Narseo Vallina-Rodriguez
Supervisor: Jon Crowcroft
Qualcomm – Cambridge
22nd May 2013
Motivation
 The web is becoming mobile
 Apps rely on multiple online/cloud services (mobile
mashup):







CDNs (Akamai)
Cloud services (Amazon WS)
Authentication APIs (Oauth)
Assisting sensors (A-GPS)
Advertisement (AdMob, Burstly, Millennial Media, …)
Push notifications (Google’s GCM)
NAT punching for P2P (Skype)
Research question
How do mobile apps’ cloud dependency
impact on cellular network and
battery life of mobile handsets?
2012-2013 outcome
 When Assistance becomes Dependence: characterizing the costs
and inefficiencies of A-GPS. Vallina-Rodriguez, Finamore,
Grunenberger, Papagiannaki and Crowcroft. ACM SIGMOBILE
MC2R (under review)
 Breaking for Commercials: Characterizing Mobile Advertising.
Vallina-Rodriguez, Finamore, Shah, Grunenberger, Haddadi,
Papagiannaki and Crowcroft. In ACM Internet Measurement
Conference 2012(IMC'12)
 Energy Management Techniques in Modern Mobile Devices.
Vallina-Rodriguez and Crowcroft. In IEEE Communications Tutorials
and Surveys, 2012.
 When David can help Goliath: the case for cellular augmentation
of wired networks. Vallina-Rodriguez, Erramilli, Grunenberger,
Gyarmati, Laoutaris, Stanojevic, Papagiannaki, In ACM HotNets'12
 Signposts: End-to-End Networking in a World of Middleboxes.
Aucinas, Chaudhry, Crowcroft, Probst Eide, Hand,
Madhavapeddy, Moore, Mortier, Rotsos and Vallina-Rodriguez. In
ACM SIGCOMM 2012. DEMO
Take away: moving to the edge!
1. Mobile applications may abuse cellular networks: they cause
network (signaling/channels/operational) and energy costs!
2. Fetching content in a centralized fashion is not the only way
Distribute as much as
you can!
Flashlinq/LTE-direct
 P2P wireless technology
 Perfect candidate for transparent communication in the edge!
 Peer discovery (energy efficient)
 Expression-based discovery (service)
 Always-on background service with low duty-cycle
 Similar to powering up a paging channel every X seconds
 Current prototype performance:
 Low-latency (<10 ms)
 Good throughput (~ 20 Mbps)
 Discovery (1~2 seconds)
… but what can be
distributed?
1. Localized data
Use case 1: Localized data
 A large fraction of mobile data is local
 Weather
 Notifications
 Ads
[Cellular data network infrastructure
characterization and implication on
mobile content placement, Xu et al.
SIGMETRICS’2011]
 Apps use cellular networks and push notifications to fetch
this content
 High latency
 No delivery guarantees
Use case 1: Airport notifications
Use case 1: Airport notifications
Traffic Pattern Heathrow App For Android (Flight Update)
Energy
Signaling
Spectrum (HSPA)
 TCP/IP Push notification model is broken for local data:
 Frequent RNC promotions (some caused by TCP Heartbeats)
 Waste of energy, middleboxes/proxies memory and radio
channels (+200K users/day, a lot of signaling traffic!)
Use case 1: Airport notifications
2. Collaborative sensing
Use case 2: Collaborative A-GPS
 Assisting data (time, ephemeris, almanac, coarse
location) downloaded from network:
 Reduces TTFF (usability)
 Temporal validity up to 2 weeks for ephemeris
 Problem: use of cellular network may impair performance
and increase energy costs!
Use case 2: Collaborative A-GPS
2x current!
Control-plane
latency
Use case 2: Collaborative A-GPS
 Collaboration between devices in a P2P fashion:
 Context-awareness (sense environment so do not turn on
AGPS indoors!)
 Share/pre-fetch assisting data (reduces latency to fetch
data)
 Prototype for Nexus One:
 Pre-fetch and cache of assisting data
 Devices can detect if they’re indoors in less than 10 seconds
 Blackbox. Hard to inject assisting data on chipsets (A-GPS is
controlled by binary/proprietary files/drivers  )
3. Wired-wireless
integration
Use case 3: Wired-wireless integration
 3G offloading to WiFi and femtocells:
 Reduce network traffic
 No real benefit for users (unless volume cap in data-plan)
 Wired network can be constrained!
 Can cellular networks augment wired networks?
 Wired nets deployment is $$$
 Cellular nets have good coverage
Use case 3: Wired-wireless integration
 Cellular network can
provide more capacity
than wired ones (DSL)
4.7 Mbps
 Spare capacity on cellular
network
 Powerboost for videostreaming apps
2 Km
2.8 Mbps
 Use-and-release
 Does NOT work
everywhere anytime!
20
Use case 3: Wired-wireless integration
 2x downlink/5x uplink
for most locations with 1
mobile device
 Simulation: 50% of the
videos have a speed
up factor of 10x
Conclusions
 Current cloud-mobile model is not efficient
 Hyper-centralized: push notifications
 Lack of connectivity between handsets: missing
opportunities
 Cellular and wired networks are fully decoupled
 Flashlinq/LTE-direct can bring a new mobile paradigm!
 Energy and network efficient
 Distributed
 Flexible
Flashlinq limitations and extensions
 Transparent security/authentication mechanisms
 Lessons to be learnt from the past: Bluetooth and WiFi-direct
failed!
 Source of DoS/Privacy/Energy attacks
 Global Signpost-ish naming (OpenSource, DNSSEC based)
 Low-level radio details must be exposed to OS!
 Too much layering hides inefficiencies: e.g. A-GPS and 3G
 Cross-layer optimizations are key (e.g. iPhone vs. Android)
 Incentives for operators?
 Reduce operational costs: better use of limited capacity
 Licensed frequency
Thank you for your attention!
http://www.cl.cam.ac.uk/~nv240
[email protected]