intrusion detection adapted for automotive challenges for ... · infineon presentation @ ieee tech...

20
Intrusion Detection Adapted for Automotive – Challenges for Hardware - An Implementation Example 2018 IEEE-SA Ethernet & IP @ Automotive Technology Day Harald Zweck, Infineon Technologies Ronny Schulze, Infineon Technologies

Upload: others

Post on 19-Jun-2020

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Intrusion Detection Adapted for Automotive – Challenges for Hardware - An Implementation Example

2018 IEEE-SA Ethernet & IP @ Automotive Technology Day Harald Zweck, Infineon Technologies Ronny Schulze, Infineon Technologies

Page 2: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018

Motivation

Source: I Am The Cavalry / Five Star Automotive Cyber Safety Framework / February 2015 / Link: https://www.iamthecavalry.org/

Statement given by “I Am The Cavalry”

Safety investigations drive substantial improvements, and records of electronic systems

operations give visibility into root causes that may otherwise be opaque. These records

can plainly show sources of error, be they malfunctions, design defects, human error or

deliberate attack. Those waiting for proof of hacking or electronic sabotage will not find

evidence without such logging and evidence collection in place. This capability will

require more effort, over time, than others on this list, but it is foundational for improving

safety in the long-term so starting now will help us achieve this goal.

Evidence Capture

Do your vehicle systems provide tamper evident, forensically-sound logging and evidence capture to facilitate

safety investigations?

I read => Electronic systems shall record events like malfunctions, defects, errors, attacks etc.

2

Page 3: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Network

Automotive Ethernet CAN / Sensor Bus

ECU ECU with Ethernet connectivity

OBD Antenna

Infineon Presentation @ IEEE Tech Day London 2018 3

Page 4: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Network Traffic Clustering

› Unexpected traffic example related to Microcontroller Units

cable defect

connector defect

unknown packet addresses

Unknown multicasts

Unused VLAN priority levels

Unexpected packets

Ethernet MAC

Microcontroller

PHY

Connector

Software Stack

Ethernet MAC

Microcontroller

PHY

Connector

Software Stack

Infineon Presentation @ IEEE Tech Day London 2018 4

Page 5: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Network Traffic Clustering

› Unexpected traffic example related to OSI Layer 1 / 2 / 2.5

MAC Address Ethertype

VLAN Tag

Physical Layer

cable defect

connector defect

unknown packet addresses

Unknown multicasts

Unused priority levels

Unexpected packets

Ethernet MAC

Microcontroller

PHY

Connector

Software Stack

Infineon Presentation @ IEEE Tech Day London 2018 5

Page 6: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Network Traffic Monitoring – Layer 1

› Layer 1 Traffic Evaluation

Infineon Presentation @ IEEE Tech Day London 2018

Front Processing

Rear Processing

Side Processing

Microcontroller

Stack

Task 1 Task 2

Stack

Ethernet MAC

500Fps

33

33

33

Fro

m T

ask 1

200Fps

30

40

30

Fro

m T

ask 2

Microcontroller

Stack

Task 1 Task 2

Stack

Ethernet MAC

Microcontroller

Stack

Task 1 Task 2

Stack

Ethernet MAC

6

1000Fps

25

50

25

To T

ask 1

500Fps

40

30

30

To T

ask 2

to / from Side Processing and Rear Processing

Page 7: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Network Traffic Monitoring – Layer 1

› Layer 1 Traffic Evaluation Example – Based on Frame Length

Infineon Presentation @ IEEE Tech Day London 2018

Front Processing

Microcontroller

Stack

Task 1 Task 2

Stack

Ethernet MAC 500Fps

40 30 30

1000Fps 25 50 25

Frame Profile (length in bytes) Task 1 reception Task 2 reception Total

Short frame

< 128

50% of 1000Fps

=> 500Fps

30% of 500Fps

=> 150Fps

500Fps + 150Fps

=> 650Fps

Mid range frame

128 < frame < 1023

25% of 1000Fps

=> 250Fps

40% of 500Fps

=> 200Fps

250Fps + 200Fps

=> 450Fps

Long frame

> 1023

25% of 1000Fps

=> 250Fps

30% of 500Fps

=> 150Fps

250Fps + 150Fps

=> 400Fps

Task 1

Task 2

7

Page 8: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

IEEE Standard Ethernet Frame

Addressing

Schemes

Link Layer

Network Layer

Transport Layer

Application Layer

Ethernet Frame

IP Packet

TCP/UDP Datagram

MAC Address Ethertype VLAN Tag TCP/UDP IP Header CRC

TCP/UDP IP Header

TCP/UDP

Session Layer

Data

(D)TLS Record

Data

Data

Data

Data

Infineon Presentation @ IEEE Tech Day London 2018 8

Page 9: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Network Traffic Monitoring – Layer 1

› Layer 1 Traffic Monitoring Example – Based on Frame Length

Infineon Presentation @ IEEE Tech Day London 2018

Microcontroller

Security Task

Security Stack

Eth

ern

et

MA

C

Qu

eu

es

Short Frame Counter

Mid Size Frame Counter

Long Frame Counter

Frame Filters

4x

3x

2x

Task 1

Stack

Task 2 Task 3

Inbound E

thern

et F

ram

es

9

Page 10: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018 10

Network Traffic Profiling – Layer 1

› Layer 1 Traffic Profiling Example – Based on Frame Length

Security Task

Security Stack

Eth

ern

et

MA

C

Short Frame Counter

Mid Size Frame Counter

Long Frame Counter

Microcontroller

time

Counter

Value Snapshot

Snapshot

Snapshot

Frame Length Profiles for Forensic Center

Page 11: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018 11

Network Traffic Profiling – Layer 1

› Layer 1 - Limitations

MAC Address

VLAN Tag

Physical Layer

Frame Layer

Packet Addresses

VLAN Tag

Priority (PCP)

Broadcast

Multicast

Frame Length

CRC Error

Covered

Not Covered

Ethernet MAC

Frame Filters

Qu

eu

es

Counters

Page 12: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Network Traffic Profiling – Layer 2

MAC Address VLAN Tag

Infineon Presentation @ IEEE Tech Day London 2018

› Layer 2 - Challenges

6 bytes = 48 bits 4 bytes = 32 bits

› 248

counters ??

› Realistic solution:

– Capture counters for passing frames -> 2 digit number range

– Capture rejected frames in one special stack process

Ethertype TCP/UDP IP Header

12

Page 13: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018 13

Network Traffic Profiling – Layer 2

› Layer 2 Ethernet Frame Profiling

Microcontroller

Eth

ern

et

MA

C

Qu

eu

es

Task 1

Stack

Task 2

Inbound Ethernet Frames

Task 3

Stack

Intrusion Detection

Stack

Address Filters

MAC Addr. Range 1 MAC Addr. Range 2 MAC Addr. Range 3 Residual Addr. Range

Page 14: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018 14

Network Traffic Profiling – Layer 2

› Multi-Layer Profiling Example

– Classify property ranges

– Capture property rate

– Count properties

MAC Address

VLAN Tag

Ethertype

TCP/UDP Port

IP Header

Count

Value Range

Rate

Intrusion Detection

Profile Values

Stack

Qu

eu

e

Address Filters

Residual Addr. Range

Task / Code Intrusion Detection

Stack

Page 15: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018 15

Simplified Network Demo

› Switched networks – Expectations

– Intelligent packet routing by switches

– Switches provide as well features for traffic analyzes

Demo Network

Device A

Device B

Device c

Switch

Page 16: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018 16

Traffic profiling/monitoring

› Using Hardware support in form of IETF MIB counter inside the Ethernet MAC

– Using register for number of good and bad packets with different length

– We count frames with a length of 64to 256 bytes, 256 – 512 bytes and 512 to 1023 bytes

› Device B and C will send there profile in a cyclic way to device A

Device A

Device B

Device C

Switch

Page 17: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Layer 2 implementation of residual Filter

› Packets which not pass the Uni/multicast addresses or VLAN filter will not be dropped

› These packets are forwarded to a residual filter queue

› Separating these traffic allows to route the traffic to a independent CPU inside the MCU to analyze

17 Infineon Presentation @ IEEE Tech Day London 2018

Eth

ern

et

MA

C

xMII

Qu

eu

es

D

MA

s

Filte

r

VLAN Tag / PCP FAIL

MAC Addresses

Resid

ual

Fil

ter

FAIL

Microcontroller

Page 18: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Separating traffics inside the MCU

Crossbar

CPU 0

RAM

Status

Control

Data

CPU x

RAM

Status

Control

Data

Eth

ern

et

MA

C

Qu

eu

es

D

MA

s

Qu

eu

es

D

MA

s

18 Infineon Presentation @ IEEE Tech Day London 2018

› Sorted packets inside the residual queue can be forwarded to an separate/isolated CPU to process the data independent

› CPU x counts periodical packets based on MAC addresses, Types etc…

Page 19: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018 19

At the End …

› See the Demo at the Infineon booth!

› What to do with all that data?

– Device B and C will report their traffic behavior to Device A which manage the switch

– Device A can analyze and may change the configuration of the switch ports for Device B and C

– Device A sends the network healthiness state to a forensic center

Page 20: Intrusion Detection Adapted for Automotive Challenges for ... · Infineon Presentation @ IEEE Tech Day London 2018 16 Traffic profiling/monitoring › Using Hardware support in form

Infineon Presentation @ IEEE Tech Day London 2018 20