Fog Computing

IEEE SDN Initiative Preindustrial Committee Workshop on Mobile Edge Cloud

With large fragmentation in the implementation of SDN/NFV, the main objective of Preindustrial committee is to create technical and economic conditions (with proper guidelines & direction) of interoperability and feasibility of technologies around SDN/NFV (widened to include the RAN) within the industry and academia using IEEE SDN initiative. In order to do this the committee will analyze and evaluate potential methodologies, make recommendations and create industrial agreements. This committee will look at opportunities and support POC around key technologies and establish relationships with other groups doing relevant work in Europe, Asia, Americas, Africa, etc. to increase collaboration & reduce duplication. (Open) Mobile Edge Cloud Draft Agenda (Subject to change based on participation)

07:00 Breakfast (60 min) 08:00 Welcome introduction by IEEE SDN Initiative 08:10 IEEE Preindustrial Committee Introduction, Objective, Background (20 min) 08:30 Organizers talks on Mobile Edge Cloud 9:30 Prof Mung Chiang (45 min) 10:15 Coffee Break (30 min) 10:45 Prof. Sachin Katti (45 min) 11:30 Prof. Guru Parulkar: ON.lab presentation on the proposed POC (45 min) 1215 Lunch (60 min) 01:15 Participant positions & presentations 03:30 Coffee Break (30 min) 03:30 Group forming & investigations, 3-4 areas, e.g.:  Service mobility/Content Distribution  Radio, core functions disaggregation  Control loops, what is centralized, what is at the edge  Mobility management 05:30 Group findings 06:30 Closing remarks 7:00 Dinner (Hosted by IEEE PC) (Open) Mobile edge cloud

Cagatay Buyukkoc, PhD IEEE SDN Initiative steering committee member Preindustrial Committee Co-chair AT&T Architecture & Design* *for Identification only, I don’t talk on behalf of AT&T

“Simplicity is a great virtue but it requires hard work to achieve it and education to appreciate it. And to make matters worse: complexity sells better.” - Edsger W. Dijkstra

“This presentation contains 'forward-looking statements' which are based on management's beliefs as well as on a number of assumptions concerning future events made by and information currently available to management. Readers are cautioned not to put undue reliance on such forward-looking statements, which are not a guarantee of performance and are subject to a number of uncertainties and other factors, many of which are outside AT&T's control, that could cause actual results to differ materially from such statements. These risk factors include the impact of increasing competition, continued capacity oversupply, regulatory uncertainty and the effects of technological substitution, among other risks. For a more detailed description of the factors that could cause such a difference, please see AT&T's 10-K, 10-Q, 8-K and other filings with the Securities and Exchange Commission. AT&T disclaims any intention or obligation to update or revise any forward-looking statements, whether as a result of new information, future events or otherwise. This information is presented solely to provide additional information to further understand the results of AT&T.”

5 5G era needs a new thinking in architecture

• Evolution & revolution in parallel • Rethink everything: Software, Automation, Shannon, Control, Cellular structure, BS, Spectrum, AI, Core, RAN, Edge, Complexity, Resilience • User centric networking • Control plane (loops) are acting in multiple time scales & in support of users & networks • Analytical framework in every aspect (optimization and more) • Address density by design, not “band-aid” solutions, how to balance with complexity?

Re: 5G PPP Horizon 2020 Use cases: (Google wouldn’t be around if use cases drove innovation) GSMA & METIS based use cases

Objective: Lets define the “edge”

Need Business & Value drivers

Intended & Uninted consequences!

• A new architectural design philosophy: • Software Defined, Virtualized & Programmable (SD/VP) ecosystem, address densification (diversity, self-backhauling, M-RAT), use distributed system principles (auto-management, resilience, software architecture,,..) and investigate techno- economical implications with key architectural principles

• NG RAN & NG Core networks are being re- architected in the industry: add edge as a new architectural component • Many new areas of applications should be enabled by the architecture Rethink Control loops

• Central control, orchestration, management, policy (ECOMP, various other COMPA, METIS MO,..) • Centralized orchestration, Service composition, Resiliency, Policy extraction (NFR->FR), coordination, Management, A-A-A • Charging, Big data analytics,

• Edge control, management & policy and Analytics • RRM* (admission control & QoS), Mobility Management , Subscriber IB proxy, Security (A-A-A), LI, Edge Analytics,

• End device control • End user participation, UCN, cost models, resiliency choices (security, performance,..), economics, more refined UE policy enfor,…SIM/UICC

• Data plane: Distributed & Programmable to a fine grain Mobility Edge efforts

• ETSI/MEC: “Mobile-edge Computing offers application developers and content providers cloud-computing capabilities and an IT service environment at the edge of the mobile network. This environment is characterized by ultra-low latency and high bandwidth as well as real-time access to radio network information that can be leveraged by applications.”

• Fog Computing: “A network architecture that uses one or more end-user clients or near-user edge devices to carry out a substantial amount of storage (rather than stored primarily in cloud data centers), communication (rather than routed over backbone networks), and control, configuration, measurement and management (rather than controlled primarily by network gateways such as those in the LTE core).”

Problem space (is same):Mobile Edge Computing proposes co-locating computing and storage resources at BS..

Solution idea (is same): MEC servers are deployed next to base stations: MEC servers are equipped with cloud HW, commodity hardware etc. and addressing transparent/pass-through and terminating applications

Components of solution:

• A virtualization layer for abstracting the cloud resources • An application platform: VM, and Application services (Traffic offload, Radio Information Services, Communication services, Registry etc • APIs to APPs

Ref: Mobile Edge Computing: A Taxonomy, Michael Till Beck, et. Al. AFIN 2014 : The Sixth International Conference on Advances in Future Internet 11 © 2013 AT&T Intellectual Property. All rights reserved. AT&T and the AT&T logo are trademarks of AT&T Intellectual Property. Fog computing/networking

Problem space (is same): Limited compute/storage on smart devices might utilize cloud computing as a extension facility;

Solution idea (is same): What if we can add: mobility support, geo- distribution, location-awareness and low latency. Fog networking/ computing is originally proposed to enable this extension at the “edge” of the network

Components of solution: (so far no architectural proposal) • Cloudlet (Satyanarayan) is a resource-rich “cloud in a box”, for use by nearby mobile devices • IOx is from Cisco works by hosting applications in a Guest Ref: Finding your Way in the Fog: Towards a Comprehensive Definition Operating System (GOS) running in a hypervisor directly on the Fog Computing, Vaquero and Rodero-Merino Connected Grid Router (CGR)

Problem space (is same): traffic between UE and services on the public cloud is “tromboned” towards the anchor point which leads to inefficiencies.

Solution idea is same: Deploying services closer to the UE, near the eNB. The services are deployed on “small scale data centers” connected to, or collocated with the eNB, called eNBC.

To make this work solve issues like, mobility management, context management etc. Excellent prospects on (O)MEC. Ref: Service Mobility in Mobile Networks, Hany Assasa, Srinivasa Vinay Yadhav and Lars Westberg Components of solution: IEEE 802.21 Media Independent Handover, Context Transfer Protocol, Session Migration Mechanism

13 © 2013 AT&T Intellectual Property. All rights reserved. AT&T and the AT&T logo are trademarks of AT&T Intellectual Property. POC with ON.lab: Mobile CORD platform (with EPC UP/CP separation) Operator’s specific Applications

Radio LTE,5G MBB Network Real-time Service Resource IoT MVNO Slices Analytics Chaining

Scheduling Enterprise Testbed …

Service View Mobile Resource & Service Orchestration CORD Platform (ONOS + XOS + OpenStack)

Networking Fabric

SGW-D PGW-D SGW-D PGW-D

Video Topology vBBU XOS SGW-C Backhaul Caching MME SGW-C View Front-haul & Optical vBBU DNS PGW-C OpenStack Backbone Switch Switch PCRF PGW-C vBBU Security ONOS CPRI (E)

Virtualized Edge Service EPC control CORD EPC control BBUs Functions Applications Platform Applications

Mobile Edge Mobile Core SGW-D, PGW-D: Data plane of SGW/PGW SGW-C/ PGW-C: Control plane of SGW/PGW (Open) Mobile Edge Cloud

• (An open) cloud platform that uses some end-user clients and located at the “mobile edge” to carry out a substantial amount of storage (rather than stored primarily in cloud data centers) and computation (including edge analytics, rather than relying on cloud data centers) in real time, communication (rather than routed over backbone networks), and control, policy and management (rather than controlled primarily by network gateways such as those in the LTE core). (Based on largely Prof Mung Chiang work)

• Develop an architecture to support this vision: • OMEC concept can be developed based on softRAN OS & fog networking amalgamation & distributed functional decomposition of both NG core and RAN functions on a common cloud platform: (MCORD or similar) • Partners: Stanford U (Prof Sachin Katti), Princeton U (Prof Mung Chiang), ON.lab, Vendor community, Chip industry, ALL! Reconstructing the network architecture: Study which ones to move to the (O)MEC ? Objective: To refactor the E2E network functions based on three industry trends: a) Re-architecting the RAN, b) Re-architecting the NG core c) (very little in devices) These two major industry efforts overlap at the mobility edge, and a new rethinking is required to address the mobility edge (and subsequently the converged wireline/wireless edge). Progress: Traditionally Core functions & RAN functions are separate. There is a lot of work on the mobile edge as a key architectural component. Several POCs are being done (here IEEE is supporting one through ON.lab. )

NG NG NG NG Core NG UE NG RAN NG Core Mobile UE RAN Edge Functions © 2013 AT&TFunctions Intellectual Property. All Functions Functions Refactor to rights reserved. AT&T and the AT&T 16 logo are trademarks of AT&T Intellectual Property. SDN based vRAN Reference Architecture (Study Areas)

Capabilities sets (DBs etc.)Subscriber Policy Charging Device

Net control& Optimization E2E Control Applications Mobility mgmt Security QoS mon. Fallout M, O, C, P MSO Nw Ctr. App Ctr. Mgm Others (OSS).

API Software Defined RAN API API Access Agnostic NG Core Network Topology and Resource Utilization RAN Apps: e.g., LTE-A, IoT, Analytics, Slicing SDN Controller

SD RAN Controller API WNOS

Data/Control Interface Access Agnostic Interface

UE/dev NGAP NGBS OMEC

D C D C D C D C P P P P P P P P

© 2013 AT&T Intellectual Property. All rights reserved. AT&T and the AT&T logo are trademarks of AT&T Intellectual Property. 17 CP/UP decoupling Decompose the layers! Mobility Control Handover forwarding Load Balancing Ciphering Admission Cntrl Segment Interference Management ReTX

Objectives: Flow QoS • Decouple CP/UP, create flexibility Management HARQ • Distribute CP/DP functions according to apps &transport options using the SD-RAN Scheduler MCS

• Create open interfaces between layers to drive innovation & insertion of “your secret sauce” Power Control MIMO control programs • Whitebox opportunities

A B C D E F G 2.46 Gbps* 720 Mbps* 360 Mbps* 180 Mbps* 27 Mbps*

RE Rx proc UE RF & A/D CP & FFT DEC MACrt MACnrt IP I/Q FW Subframe Rx data Equal., MIMO SoftBit MAC MAC PDCP demap rec IDFT, demap FW FW FW payload payload payload

RRU BBU

L1 L1’ L1”” L2-RT L2-NRT L3

• RLC CP DP CP DP CP DP • PDCP CP DP CP DP CP DP • RLC • Scheduler FPGA/ASIC/RF DSP/HW Acc./ASIC Specialized/but opening will drive standardization/price drop 18 © 2014 AT&T Intellectual Property. All rights reserved. AT&T and the AT&T logo are trademarks of AT&T Intellectual Property. CPU + NPU AT&T Proprietary (Internal Use Only) Not for use or disclosure outside the AT&T companies except under written agreement.

*Example : 20 Mhz/2x2 antenna config. MCS of Spectral eff=3 bit/cu E2E Software Defined Architecture OTT MVN Enterprise Value addedO Service Chains （Firewall,NAT，…) Integrated Cloud API Infrastructure Access diversity Big Data Capability sets Value added Devices MANO DB’s, profiles, Services device man, SDN controller charging,… Areas of Refactored NGxCore-c Investigation Refactored RAN-c

NG AP Open Mobile Edge Cloud Refactored MANO CORE-e Coordination Refactored E2E Architectural principles Flexible transport options RAN-e Wi-Fi (FH/MH/BH) • Simplicity (as opposed to complexity SD RAN Resource CONTROLLER • Efficiency (Energy & spectrum), Modularity, Scalability, Flexibility

Data AP Data/Control • Resiliency Interface • Software architecture focus • Shared Cloud platform Environments • Drives an (Open) Mobile Ecosystem • Compatible with SDN/NFV frameworks that are Programmable • Good TCO/flexible transport • Future proof © 2013 AT&T Intellectual Property. All rights reserved. AT&T and the AT&T logo are trademarks of AT&T Intellectual Property. 19 • Allows (Easily deployable) Value Creation platforms Summary & next steps

5G era is more than an air interface and would require rethinking everything!

No buzzwords: Use a word means developing a capability to measure, analyze!

Mobile edge is a key intersection where 5G will be based

Next steps: Create a common architecture across the industry, unify the approaches, based on initial decompositions to see what additional POCs to schedule in 2016+.

Need to merge different methodologies in one common architecture: use your secret sauce on it!