placing content intelligence at 5g mobile network …...2015/11/03 · mobile edge computing (mec)...
TRANSCRIPT
Friday, 30 October 2015 1
Placing Content Intelligence at 5G Mobile Network Edge
Ning Wang
University of Surrey
Confidential – for 5GIC Members Only
5G Innovation Centre (5GIC) Vision on 5G
Wednesday, 04 November 2015 2
Mobile application content/data+
Context information
“User perceived infinite bandwidth capacity”
5GIC’s two complementary solutions:
• Efficient use of radio spectrum
• Intelligent network handling of data
DATA
Confidential – for 5GIC Members Only
Our Research Landscape
Wednesday, 04 November 2015 3
Context awareness
Content/context aware 5G network architecture
Network support
Context (knowledge) input
User context
Co
nte
xt
da
ta e
mb
edd
ing
Content context
Applications
Optimisationtargets
• Multimedia content• Internet of things (IoT)
• Enhanced User experiences
• Network efficiency• Device efficiency
Network functions• Discovery / retrieval• Caching / pushing / adaptation• Routing /forwarding• Context data analytics
Network context
Device context
Confidential – for 5GIC Members Only
Intelligence on Video Content Handling: Network edge vs. Devices
Wednesday, 04 November 2015 4
Content source
(DASH server)
Cached DASH segments with
different resolution levels
Mobile
network
Knowledge on:
Content, users, devices
and network conditions
OTT delivery
Context information on:
Smart decision making by 5G devices – adaptively control video content downloading across parallel available RATS, based on the provided network context
Smart decision making by 5G Network, based on Mobile Edge Computing (MEC) –content caching and active pushing towards UEs, based on the provided device context (e.g. buffer level)
Network context
User/device
context
Confidential – for 5GIC Members Only
Mobile Edge Computing (MEC) Enablement
Wednesday, 04 November 2015 5
Goals
To leverage MEC as a key enabler for supporting smart network
operations in different mobile applications
MEC functions
Video content caching
Video adaptation and pushing
Enables
Metadata caching and processing for
web content
Proximity service control for D2D
networking
Minimise video access delay
Enhance video quality
Minimise web downloading
time
Smart not-spot coverage
Bring content close to users
Reduced RTT increases
TCP throughput
Actively feed UE with
optimised resolution levels
Connection optimisation based
on cached metadata
User mobility tracking and
prediction
Confidential – for 5GIC Members OnlyWednesday, 04 November 2015 6
Device-to-Device Networking in 5G
• Background: devices as part of 5G network
• Use case 1: Not-spot network coverage through D2D relay:– A “foreign” device makes a request for network access through D2D links
and local devices in proximity become helper candidates– Supported with decision-making algorithms on helper selection by
the client according to specific requirements and context information (e.g. client/helper mobility)
• Use case 2: D2D content caching: Pushing information centric networking (ICN) to the device level– Popular mobile content objects are pre-loaded a priori to a (small) set of
helper candidates to be disseminated to encounter clients in a opportunistic way
– Content requests originated from clients are sent to the ICN-supported network which is able to direct requests to the in-proximity helpers holding the content
– Supported with optimisations on (offline) helper selections and (online) content pushing/replacement at the helper device level
Confidential – for 5GIC Members OnlyWednesday, 04 November 2015 7
How MEC Plays a Role?
• User (helper) mobility – a double-edged sword
– Offers opportunities to get clients connected through D2D
– Causes connection disruptions/churns due to ad hoc departure
• Network’s knowledge on local mobile user mobility
– Capability of predicting the mobility behaviours of local connected users as potential helpers
– Practicality considerations: accuracy of human mobility prediction
• 80% - 93% spatiotemporal predictability of human mobility according to [1]
• Network’s intelligence
– Capability of computing the best combination of helpers in order to satisfy clients requirements, in particular towards minimised disruption
[1] Chaoming Song et al, “Limits of Predictability in Human Mobility”, Science, Vol. 327, 2010, pp. 1018-1-21
Confidential – for 5GIC Members OnlyWednesday, 04 November 2015 8
Summary
• In order to achieve “user perceived infinite capacity”, smart network functions need to be in place for complementing the radio technologies
• At 5GIC We envisage a wide variety of mobile services that can potentially benefit from MEC
– Content based: including both video- and web-based applications
– Non-content based: e.g. network-assisted D2D networking
• What can be smart network functions offered from MEC?
– Direct involvement of handling data/content in the user plane: e.g. caching, adaptation, pushing, prefetching etc…
– Smart control/coordination based on locally maintained context knowledge: e.g. D2D helper identification