Securing the Software Defined Car™ Using Artificial Intelligence and OTA Updates

1www.movimentogroup.com

Securing the Software Defined Car™

Using Artificial Intelligence and OTA Updates

November 14, 2017

Live Webinar Series


Industry Speaker Host Presenter

Follow me on

Mahbubul AlamCTO & CMOMovimento Group

Follow me on

Sushaiv WadekarAVP, Automotive ERSHCL Technologies

Mahbubul Alam is an international entrepreneur, technology and business executive with broad management experience. A frequent author, speaker and multiple patent holder, he currently holds the position of CTO/CMO at Movimento, a Delphi company, reinventing the company’s technology and strategy. He has been honored with the ‘2016 Bay Area CIO of the Year’ award finalist by the Silicon Valley Business Journal and the San Francisco Business Times.

Highly accomplished and successful automotive advanced technology executive in sales & business development. Over 25 years of high level, senior executive experience with OEM, tier one, tier two automotive direct sales, as well as distribution and sales rep management for corporate sales revenue and margin growth.


Agenda

Ø State of Security

Ø Cyber Risks

Ø E2E Security Architecture

Ø Security Best Practice

Ø Introduction to AI

Ø Proactive OTA Update

Ø Key Takeaways


State of The Security SolutionsSolutions since 2000 Challenges in 2017 ü Firewall

ü IDS/IPS

ü Threat Intelligence

ü Data Loss Prevention (DLP)

ü Security Info & Event

Management (SIEM)

ü Vulnerability Scanner

ü Security Analysis

ü Auditing Still a Challenge

ü Still Too Many False Positives

ü Advanced IDS Signature “False Positive”

ü Another IDS Offering

ü Challenge with Parsing, Context

and Prioritization

ü Same Network or Apps Interface

ü SIEM Can Be Used For It


Automotive Cybersecurity Risks

Attack Vectors Potential Impacts

ü Phishing

ü DNS Spoofing

ü Denial of Service

ü Laterally Moving Malware

ü Location Spoofing

ü OTA Re-Tasking

ü Credential/Data Theft

ü Poor Operation

ü No Operation

ü Ransom or Hijack

ü Vehicle Theft

ü Terrorism


5Cs of the On and Off Board Secure Car Concept

Chipset

Client

Connectivity (On/Off Board)

Cloud/FOG Computing

Content

TPM/HSM

CodeObfuscation

Transport&SessionSecurity

SecureVirtualInfrastructure

SecureApplication


Software Defined Car: 4 Layer Security StackSecure Application & Update

• Maps, Media, Info, …

• User Profile, Productivity, …

Secure AI, -Cognitive Systems & Update

• Algorithm, 3D HD Maps, …

• Image Recognition, …

Secure Software & Update

• Gateway, Domain Controller, …

• OS (ADAS, LIDAR, IVI, …)

Secure Reconfigurable HW & Update

• ECU, Sensor Fusion ECU, …

• TCU, Module, …


E2E SDC Security: Closed Loop Architecture

MovimentoClient

ContentDelivery

InfotainmentService

App Store

SoftwareUpdates

SecureOTA/DataClient

OEMs

Tier-1Suppliers

ServiceProviders

ContentProviders

SecureOver-The-AirDelivery

SecureOver-The-AirData

MovimentoUnifiedOTACloudPlatform

App Delivery

OTASoftwareUpdates

InfotainmentService

Management ContentDelivery

SoftwareManagement

Big Data

CyberSecurity

SDP


Categories of Artificial Intelligence

Artificial

Narrow

Intelligence

Artificial

General

Intelligence

Artificial

Super

Intelligence

Machine Intelligence For Specific Task >= Human Efficiency

Machine Intelligence For Any Task >= Human Intelligence

Machine Intelligence For Every Field > Best Human Intelligence


Machine LearningAbility to learn without being explicitly

programmed

Deep LearningExceptionally effective

at learning patterns

Neural NetworksDerives meaning from data using

hierarchies and layers “mimicking the Neural Networks of our brain”

Artificial Intelligence Family

DL

NN

ML

AI


Machine Learning

Data Labeled Data

Learned Model Algorithm

Prediction

TrainingPrediction


Types of Machine Learning

Supervised

Learning

Unsupervised

Learning

Reinforcement

Learning

Learning with a Labeled Training Set

Discovers Patterns in Unlabeled Data

Learning Based on Feedback or Rewards


Supervised & Unsupervised Machine Learning

Supervised - Predictive Unsupervised - Descriptive

Classification

Document

Category(A,B,C,D..)

Regression

Clustering

Anomaly

Detection


Deep Learning

A neural network consists of

a hierarchy of layers where

each layer transforms the

input data into more

abstract representation.


Machine vs. Deep Learning


Advanced Security Concepts

Continuous Authentication Machine Learning for User Behavior Analysis

Micro Segmentation Secure Enclaves with Software Defined Perimeter

Dynamic Policy Decision Automation with Closed Loop Architecture

Threat Intelligence Real-time Sharing & Crowd Sourcing


AI Based Threat Prediction & Counter Measure

Historical Data(Communication)

SupervisedLearning

Breached SecurityDetected Anomaly Claimed Fraud

Prediction

LearnedModel

Real-timeCommunication Data

Security Threat Monitoring &

Proactive Counter

Measurement System

Proactive Security OTA Update

Security BreachAnomaly PredictedFraud Claim

Training Prediction OTA Update


Key Takeaways

Closed Loop Architecture

For E2E security, a closed loop

architecture is a must inside and

outside of a vehicle for

operations & smart mobility

services.

Connected Services

Secure enclaves for OEMs, Tier-1s and ecosystem

partner applications and

services will be key to the SDC

era.

Technology & Timing

Securing connected cars

does not guarantee success for car sharing and autonomous cars.Proactive security counter measures

are needed.

New Business ModelFor fully

autonomous cars and even for

car/ride sharing economy, human

safety, security, stress free and

trusted platform are required.

In-Vehicle ArchitectureThe 5C security concept for on and off board

vehicle is vital. In-vehicle 4 layer security and enclaves are must-haves.


Q & A


Saving Lives Using Artificial Intelligence and

Context-based Automotive OTA Software Updates

Date: Tuesday | 12 December, 2017Time: 8 AM PST / 11 AM EST / 5 PM CET / 8:30 PM IST

Upcoming Webinar


www.movimentogroup.com

www.linkedin.com/company/movimento

@movimentoauto