5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom...

25
5G and Privacy - Diversification of Services and Tenants - Dr. Artur Hecker Director Future Network Technologies Munich Research Center Huawei Technologies, Germany Contact: [email protected]

Transcript of 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom...

Page 1: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

5G and Privacy

- Diversification of Services and Tenants -

Dr. Artur HeckerDirector Future Network Technologies

Munich Research CenterHuawei Technologies, GermanyContact: [email protected]

Page 2: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Preliminary thoughts on 5G and privacy aspects

Why? Most of real IoT today is cellular

Mostly seen from the Core network research perspective

The opinions expressed herein are strictly personal

About this talk

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 2

Page 3: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Public Land Mobile Networks

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 3

Also called Cellular or Mobile Networks

Success story started in 1992 with GSM (so-called 2nd Generation, 2G)

10 year generation cycles in definition, standards, R&D, deployment, etc.

1990: GSM; 2000: 3G; 2010: 4G, 2020: 5G

Since mi 90s: by 3GPP

3GPP output becomes ETSI norms

Tremendously successful

Source: GSMA, https://www.gsma.com/mobileeconomy/wp-content/uploads/2018/05/The-Mobile-Economy-2018.pdf

WORLD GLOBAL

Page 4: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

PLMN: Basic Principles

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 4

Digital radio, hierarchical telco system

Service for mobile equipment from terrestrial radio towers organized through a core network

UE | RAN | CORE; separate user, control and management planes

Support for UE mobility with session continuity (mainly MBB service)

Requires continuous awareness of the “network” about the location of the UE

Since 2G: call from the train or car. Since 4G: “always on” Internet access service

Integrated subscriber management

Subscriber = IMSI (international mobile subscriber identity, “+49 171 12345678”)

Links contract (Mr. John Doe, payment) and technical provisions (authorizations, service types, …)

Philosophy

Operators are trustworthy, local States have the jurisdiction, technology to be adapted accordingly

Page 5: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Development of PLMN

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 5

Source: GSMA, https://www.gsma.com/mobileeconomy/wp-content/uploads/2018/05/The-Mobile-Economy-2018.pdf

Page 6: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Personal Data Protection – Regulatory Aspects

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 6

Operation subject to restrictions on the user data usage

Directive 95/46/EC (data protection)

Directive 2002/58/EC (privacy)

Directive 99/5/EC (radio equipment)

ETSI TR 187 010 - identity management

Other regulations (anti-terrorist laws, etc.)

No contract without a valid ID presentation, i.e. no anonymous SIM cards

Lawful interception

Special exceptions for disclosing of PII with law enforcement agencies: obligatory for service providers

Newer regulations

eIDAS Regulation, GDPR: establish trust in the Digital Single Market

- legal obligations to preserve a user's control of their identity in electronic communication

- obligations intended to avoid frauds

Page 7: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Data fusion and re-identification

Data breaches

Service termination / inaccessibility

Lock-in mechanisms

Malware

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 7

Over-collection

Mis-contextualization

User impersonation

Alteration of ownership / access

rights

Alteration of persistence

Threats to Personal Data

Cmp. ETSI TR 103 304 V1.1.1

Page 8: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Protection of Personal Data – Technology Aspects

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 8

Generalized cryptography usage (depending on the local jurisdiction)

Subscriber authentication using SIM card and the HSS (since 3G: mutual)

Enciphering of user payload on the radio link (since 3G: enciphering and integrity protection)

IT engineering for system protection

Pseudonyms on recurrent operations (a so-called TMSI can be used instead of IMSI)

Access controls within the control and management planes

Lawful interception

Standardized LI solutions for voice, conferencing, IMS-based services, messaging (SMS, e-mail, etc), and

Internet access

Known weak points

Roaming scenarios, SMS, internal signaling protocols, external correlation of system identifiers

Page 9: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

4G: Implementation Architecture

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 9

Source: “Survey on Threats and Attacks on Mobile Networks”, IEEE Access, vol 4, August 2016

A system engineered to

provide mobile service to

subscribers

Subscribers: humans with handhelds

Target: always on data service

The composition of system

entities realizes the service

Reliable yet rigid

Options are contradictory

Page 10: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

• 1 ms E2E latency (radio)

• 10Gbps per connection

• 1,000K connections per km2

• 500km/h high-speed

• Slicing network architecture

Goal: One Network Fits All Applications with High Flexibility

Ultra High Reliability Ultra Low LatencyUltra High Throughput

Smart Manufacturing Connected CarAugmented Reality

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 10

5G Use Cases

Page 11: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

5G: Key Requirements from Business

From Huawei WP

Integration of vertical industries (Mission-)critical services

mMTC, uRLL, eMBB

Diversity of services Rapid service deployment

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 11

Page 12: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

5G Slicing: Common Understanding

Definition: 5G E2E Network Slicing is a concept for running multiple logical networks (which could be customized and with guaranteed SLA)

as virtually independent business operations on a common physical infrastructure.

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 16

Page 13: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Potential Business Cases Cloud and Connectivity are the dual engines of digital transformation.

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 17

Page 14: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Logical Architecture for V2V Mobility

eNB

MME’

HSS

Logical Architecture

RAN CORE

MME, SGW

PGW

RU, DU

PGWSGW

MME

eNB

PCRF

HSS

GTP-U

GTP-C

Physical Network

IP backhaul

4G: single, unique LTE EPC

Logical Architecture for legacy MBB

RAN CORE

COTS Platform

Data Center

RU, DU PGWSGW

MME

eNB

PCRF

HSS

GTP-U

GTP-C

Physical Network

IP backhaul

5G, with slicing

Logical Architecture for MTC

vPGWeNB

PCRF

HSS

Slicing produces multiple versions of all instancesExamples (for cars, sensors, mobiles):- Mobility handlers- Connectivity handlers- Authentication handlers- Session handlers

Yet, many of these instances are similarAtomic FunctionsFunction Composition

Slicing and Modular Architecture

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 18

Page 15: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

From 3GPP TR23.799Key issue 1: Support of network slicingKey issue 2: QoS frameworkKey Issue 3: Mobility management frameworkKey issue 4: Session managementKey issue 5: Enabling (re)selection of efficient user plane paths Key issue 6: Support for session and service continuityKey issue 7: Network function granularity and interactions between them Key issue 8: Next Generation core and access - functional division and interfaceKey Issue 9: 3GPP architecture impacts to support network capability exposure Key issue 10: Policy FrameworkKey issue 11: ChargingKey issue 12: Security frameworkKey issue 13: Broadcast/Multicast CapabilitiesKey Issue 14: Support for Off-Network CommunicationKey Issue 15: NextGen core support for IMSKey Issue 16: 3GPP system aspects to support the connectivity of remote UEs via relay UEs Key Issue 17: 3GPP architecture impacts to support network discovery and selectionKey Issue 18: Interworking and MigrationKey Issue 19: Architecture impacts when using virtual environmentsKey issue 20: Traffic Steering, Switching and Splitting between 3GPP and non-3GPP AccessesKey Issue 21: Minimal connectivity within extreme rural deployments

In red: priority key issues

The general direction is being standardized by 3GPP

3GPP: Slicing and atomic functions in standardization

3GPP Release 15 Working Model: TS23.501Modularization as key design elementAlso: Service-Based Architecture

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 19

NGMN: slice templates as representations of long term business contracts

Page 16: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

First 5G-relevant release, on time but mainly MBB-oriented

Defines New Radio (NR) and 5GC (5G Core Architecture)

Core Network Architecture with SBA: TS 23.501, TS 23.502, TS 29.500 ready

Management plane work still ongoing (SA5, 28.xxx series)

Rel 16 expected to sharpen

URLLC work

mMTC work

V2X

NWDA

eSBA

Many procedures in general

3GPP Release 15

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 20

Page 17: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Drivers/enablers of the transformation

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 21

The magic triangle for realizing slicing:

Orchestration (mgmt)(MANO, Heat, ZooKeeper, VNE, …)

Software-Defined

Networking(SDN, ForCES,

I2RS) Network Function Virtualization

(NFV, MEC)

SDN ControllerVirtual

Infrastructure Manager

NFVSDN

NE1 CEkNEk CE1 CE2

Orchestrator

Current Reference Model

Problematic:- Artificially classifies network functions- Single entity capacity restrictions/limits- Failure treatment- Distributed network functions

Page 18: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

S1S2 S3

S4C1

C2

C3

C4

S5S6

NF4

NF2

NF3

Continuous creation and removal of numerous components

Paths and end-points are part of the potentiallycritical internal network function realization

NF1

Guaranteed, correct execution of all this is required:- Correct local execution of hundreds of thousands of components

- Shared with other executed components, yet guaranteed – scheduling

- Correct, guaranteed interconnection of components in spite of contention- Timely control of all this

- For creation/migration, path QoS, execution guarantees, checks, quotas, …

- Correct function of the executing environment- Including all resources (compute, network)- … and their control systems

5G requires more than best effortbecause of its own realization

Slicing: from HW Composition to SW CompositionPlethora of modules/paths - must work correctly any time

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 22

Page 19: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

5G: A metaphor

Ready to use• Can type a text straightawayHardware defines the service (1 service)• Hardware limitations = service limits

Not ready to use• Need to install and start MS Word first (mgmt)• Need to execute MS Word (runtime control)

• Runtime resource allocation (CPU, memory, disk)• In parallel to other applications

Software defines the service (N services)• Hardware limitations = service quality limits

4G 5G

NFV/SDN

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 23

Page 20: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

#Slices X (#NF types per slice) X (#NF instances)

~1000 X ~10 ~10 = 105X

We need a lot of automation, i.e. control!

e.g. according to NGMN, operators cf. 3GPP NextGen SA2, RAN3

Number of Paths between modules: worst case scales in O(#NFI2)ϵ [106 ; 1010]

Example. Depends on type and load.e.g. how to serve billions of IoT/M2M devices?

of NF Instancesrunning in the network

Number of modules (#NFI) =

On the expected scalesNumber of modules, number of paths

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 24

Page 21: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 25

Possible future of mobile systems:An on-demand interconnection of many virtual and physical entities

Page 22: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

5G: changes on the horizon

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 26

Multiple Tenants / Multiple Slices

Advent of business users in cellular systems

More and more diversified subscribers

Potential to a more complex realization

IT/telco, virtual-physical, cloud/network, small cells

More modules/functions, more flavors, more instances

Virtualization: unrelated, unknown, shared platform usage. Curious cloud providers

Function-external user state (e.g. SBA SMF accessing external USM data)

“Differently from previous telecom scenario where user data was mostly accessible from network functional elements, several

kinds of information are today easily accessible from terminal equipments or end user devices, through open and specialized

Application Programming Interface (API). Thus, it may be difficult to have a priori knowledge of who may need access to users'

data, when and where this may happen and whether that data could be or contain PII” (from ETSI TR 103 304 V1.1.1)

Page 23: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Slicing and network softwarization are new challenges in networking

5G is all about vertical businesses

Potential new privacy problems due to a more complex environment

Multi-tenancy by design

New service types – new data exchanges and flows, new PII => requires new protective measures

Virtualization / external platforms

New market actors

New markets for operators, with different regulation (eHealth)

Conclusions

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 27

Page 24: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,
Page 25: 5G and Privacy - Webster University Geneva · 2019-02-08 · “Differently from previous telecom scenario where user data was mostly accessible from network functional elements,

Normative References

A. Hecker | Geneva, Switzerland | June 1, 2018 5G and Privacy 29

3GPP TS 22.261 (requirements)

3GPP TS 23.501 (system architecture)

3GPP TS 23.502 (core network procedures)

3GPP TS 29.500 (implementation aspects)

ETSI TR 103 304 V1.1.1 (2016-07)

http://www.etsi.org/deliver/etsi_tr/103300_103399/103304/01.01.01_60/tr_103304v010101p.pdf

ETSI TR 101 567