LTE security and protocol exploits

Roger Piqueras Jover

Wireless Security Research Scientist – Security Architecture – Bloomberg LP

ShmooCon – January 2016

About me
• Wireless Security Researcher (aka Security Architect) at Bloomberg LP
– http://www.bloomberg.com/company/announcements/mobile-security-a-conversation-with-roger-piqueras-jover/
• Formerly (5 years) Principal Member of Technical Staff at AT&T Security Research
– http://src.att.com/projects/index.html
• Mobile/wireless network security research
– LTE security and protocol exploits
– Advanced radio jamming
– Control plane signaling scalability in mobile networks
– 5G mobile networks and new mobile core architectures
• If it communicates wirelessly, I am interested in its security
– Bluetooth and BLE
– 802.11
– Zigbee, Zigwave
– LoRa, SIgFox…
• More details
– http://www.ee.columbia.edu/~roger/ - @rgoestotheshows
2
Mobile network security
• Often thought at the “app” layer • My areas of interest
– Certificates – PHY layer
– Encryption – “Layer 2” protocols (RRC, NAS, etc)
– SSL – Circuit-switched mobile core architecture for
– Recent examples packet-switched traffic  No bueno!
• iOS SSL bug – Recent examples
• Android malware • LTE jamming
• XcodeGhost iOS infected apps • Low-cost LTE IMSI catchers and protocol exploits
• Long etc • IM app causes huge mobile operators outage
• Mobile operators trouble with “signaling storms”

3
Mobile network security

The first mobile networks were not designed with a strong security focus (no support for encryption in 1G!!!)

“Old” encryption Strong encryption Stronger encryption

Device Mutual Mutual
authentication authentication authentication

Basic security principles

• Confidentiality Protecting user data
• Authentication
Mobile connectivity availability
• Availability
against security threats

4
 LTE basics…

5
LTE mobile network architecture

6
 Decode PSS and SSS to synchronize in
LTE Cell Selection and Connection time and frequency.

Obtain
Cell Search
System Decode PBCH Power up
Configuration Procedure

RACH
Idle

Random Radio Access

Access
Connected User traffic
Bearer + (Attach)

• System configuration
– Decode Master Information Block (MIB) from PBCH
– Decode System Information Blocks (SIBs) from PDSCH
8
LTE frame

9
LTE NAS Attach procedure

10
Mobile network user/device identifiers

IMEI – “Serial number” of the device

IMSI – secret id of the SIM that should never be disclosed

TMSI – temporary id used by the network once it knows who you are

MSISDN – Your phone number.

XYZ-867-5309

11
 LTE security and protocol exploits…

13
LTE security and protocol exploits
• Sniffing base station and network configuration broadcast messages
• LTE security
• LTE IMSI catchers
• Mapping of {phone number, TMSI, IMSI}
• Bricking/blocking devices and SIMs
• LTE location leaks and tracking target devices

14
Sniffing base station configuration

Time: 00:02:10.087204 Frame: 93

Subframe: 0
BCCH-BCH-Message
message
dl-Bandwidth: n50
phich-Config
phich-Duration: normal
phich-Resource: one
systemFrameNumber: {8
bits|0x17}
spare: {10 bits|0x0000|Right
Aligned}

LTE PBCH MIB packet

15
Sniffing base station configuration
Time: 00:02:10.102204 Frame: 94 Subframe: 5
BCCH-DL-SCH-Message
message
c1
systemInformationBlockType1
cellAccessRelatedInfo
plmn-IdentityList
PLMN-IdentityInfo
plmn-Identity
mcc
MCC-MNC-Digit: 3
MCC-MNC-Digit: 1
MCC-MNC-Digit: 0 Mobile operator
mnc
MCC-MNC-Digit: 4
MCC-MNC-Digit: 1
MCC-MNC-Digit: 0
cellReservedForOperatorUse: reserved
trackingAreaCode: {16 bits|0x2713}
cellIdentity: {28 bits|0x0075400F|Right Aligned} Cell ID
cellBarred: notBarred
intraFreqReselection: allowed
csg-Indication: false
cellSelectionInfo
q-RxLevMin: -60
freqBandIndicator: 17
RX power to select
schedulingInfoList that cell
SchedulingInfo
si-Periodicity: rf8
sib-MappingInfo
SIB-Type: sibType3
si-WindowLength: ms10
systemInfoValueTag: 11
Padding

16
Sniffing base station configuration

RACH config

Etc…RRC timers
Paging config

User traffic
config

17
LTE PDSCH SIB2/3 packet
Sniffing base station configuration
• MIB/SIB messages are necessary for the operation of the network
– Some things must be sent in the clear (i.e. a device connecting for the first time)
– But perhaps not everything
• Things an attacker can learn from MIB and SIB messages
– Optimal tx power for a rogue base station (no need to set up your USRP to its max tx power)
– High priority frequencies to force priority cell reselection
– Mobile operator who owns that tower
– Tracking Area of the legitimate cell (use a different one in your rogue eNodeB to force TAU update messages)
– Mapping of signaling channels
– Paging channel mapping and paging configuration
– Etc

18
LTE security
RACH handshake
between UE and eNB
RRC handshake between
UE and eNB

Connection setup
(authentication, set-up of
encryption, tunnel set-up,
etc)

Encrypted traffic

19
LTE security

Unencrypted and unprotected. I can

sniff these messages and I can
transmit them pretending to be a
legitimate base station.

Other things sent in the clear:

• Measurement reports
• Measurement report requests
• (Sometimes) GPS coordinates
• HO related messages
• Paging messages
• Etc

20
LTE security

Regardless of mutual authentication and strong encryption, a mobile device engages in a

substantial exchange of unprotected messages with *any* LTE base station (malicious or
not) that advertises itself with the right broadcast information.

21
LTE open source implementations
• There are a few somewhat fully functional LTE open source implementations
– OpenLTE – End to end implementation: RAN and “EPC”.
• http://sourceforge.net/projects/openlte/
– gr-LTE – Based on gnuradio-companion. Great for people new to software radio.
• https://github.com/kit-cel/gr-lte
– OpenAirInterface – Industry/Academia consortium.
• http://www.openairinterface.org/
– srsLTE – Almost complete implementation. Includes srsUE, device open source implementation.
• https://github.com/srsLTE
• Hardware setup
– USRP B210 for active rogue base station
– BUDGET: USRP B210 ($1100) + GPSDO ($625) + LTE Antenna (2x$30) = $1785
– Machine running Ubunutu
– US dongles (hackRF, etc) for passive sniffing.

All LTE active radio experiments MUST be performed inside a faraday cage.

22
LTE traffic captures
• Sanjole WaveJudge 5000 with IntelliJudge traffic processor
– Reception and sniffing from multiple eNBs simultaneously
– Decoding of messages at very low SNR regime
– Retransmission of captures
– Thanks to Sanjole for helping out and providing all the captures shown in this presentation!
– http://www.sanjole.com/our-products/lte-analyzer/

• Other options
– openLTE pcap traffic dump
– WireShark LTE RRC library
– hackRF
– Other LTE open source implementations

23
LTE IMSI catcher (Stingray)
• Despite common assumptions, in LTE the IMSI is always transmitted in the clear at least once
– If the network has never seen that UE, it must use the IMSI to claim its identity
– A UE will trust *any* eNodeB that claims it has never seen that device (pre-authentication messages)
– IMSI can also be transmitted in the clear in error recovery situations (very rare)
• Implementation
– USRP B210 + Ubuntu 14.10 + gnuradio 3.7.2
– LTE base station – OpenLTE’s LTE_fdd_eNodeB (slightly modified)
• Added feature to record IMSI from Attach Request messages
– Send attach reject after IMSI collection

• Stingrays also possible in LTE without need to downgrade connection to GSM

24
LTE IMSI catcher

25
LTE IMSI catcher

26
Mapping {phone number, TMSI, IMSI}
• Given a phone number
– Paging messages broadcasted in the clear and addressed to the TMSI
– Silent text messages to target device
• Setting up a rogue base station
– UE will attempt first with TMSI
– Then intercept IMSI
• Cool new tricks in a paper I will discuss shortly…

27
Mapping {phone number, TMSI, IMSI}

28
(Intermission) - Some excellent related work
• A team at TU Berlin, University of Helsinki and Aalto University doing excellent work in the same area
– More results on SIM/device bricking with Attach/TAU reject messages
– LTE location leaks
– Detailed implementation and results
– Paper to be presented at NDSS: http://arxiv.org/abs/1510.07563

• Prof. Seifert’s team at TU Berlin responsible for other previous VERY COOL projects
– Respond to phone calls and receive text messages that are intended for somebody else (USENIX 2013)
– Preventing signaling-based attacks coming from smartphones (IEEE DSN 2012)
– SMS baseband fuzzing (USENIX 2011)
– Mobile botnets (MALWARE 2010)

• The authors have submitted their Wireshark LTE dissectors and are being merged into the application
– Really looking forward to this…

29
Device and SIM temporary block
• Attach reject and TAU (Tracking Area Update) reject messages not encrypted/integrity-protected

• Spoofing this messages one can trick a device to

– Believe it is not allowed to connect to the network (blocked)
– Believe it is supposed to downgrade to or only allowed to connect to GSM

• Attack set-up
– USRP + openLTE LTE_fdd_eNodeB (slightly modified)
– Devices attempt to attach (Attach Request, TAU request, etc)
– Always reply to Request with Reject message
– Experiment with “EMM Reject causes” defined by 3GPP Real eNodeB

These are not the droids we are looking

for. I am not allowed to connect to my
provider anymore, I won’t try again.
REQUEST

30
REJECT
These are not the droids you are looking for… And you are not Rogue eNodeB
allowed to connect anymore to this network.
Device and SIM temporary block
• Some results
– The blocking of the device/SIM is only temporary
– Device won’t connect until rebooted
– SIM won’t connect until reboot
– SIM/device bricked until timer T3245 expires (24 to 48 hours!)
– Downgrade device to GSM and get it to connect to a rogue BS

• If the target is an M2M device, it could be a semi-persistent attack

– Reboot M2M device remotely?
– Send a technician to reset SIM?
– Or just wait 48 hours for your M2M device to come back online…

31
Soft downgrade to GSM
• Use similar techniques to “instruct” the phone to downgrade to GSM
– Only GSM services allowed OR LTE and 3G not allowed

• Once at GSM, the phone to connects to your rogue base station

– Bruteforce the encryption
– Listen to phone calls, read text messages
– Man in the Middle
– A long list of other bad things…

(Much more dangerous)

I will remove these restraints and rogue GSM base station
leave this cell with the door open…
and use only GSM from now on…
and I’ll drop my weapon.
REQUEST

32
REJECT
You will remove these restraints and leave this cell with the Rogue eNodeB
door open… and use only GSM from now on.
LTE location leaks and potential target device tracking
• RNTI
– PHY layer id sent in the clear in EVERY SINGLE packet, both UL and DL
– Identifies uniquely every UE within a cell
– Changes infrequently
• Based on several captures in the NYC and Honolulu areas
• No distinguishable behavior per operator or per base station manufacturer
– Assigned by the network in the MAC RAR response to the RACH preamble

33
LTE location leaks and potential target device tracking

34
LTE location leaks and potential target device tracking

35
LTE location leaks and potential target device tracking
• Potential RNTI tracking use cases
– Know how long you stay at a given location
• and meanwhile someone robs your house…
– Estimate the UL and DL load of a given device
• Signaling traffic on the air interface << Data traffic on the air interface
– Potentially identify the hot-spot/access point in an LTE-based ad-hoc network

• Phone # - TMSI – RNTI mapping is trivial

– If the passive sniffer is within the same cell/sector as the target

36
 Handoff between cell 60 and cell 50

37
Cell ID = 50
Cell ID = 60
 Handoff between cell 60 and cell 50

38
 Handoff between cell 60 and cell 50

39
 Handoff between cell 60 and cell 50

40
 Handoff between cell 60 and cell 50

41
0x2A60 = 10848
 Handoff between cell 60 and cell 50

42
0x2A60 = 10848
 Handoff between cell 60 and cell 50

43
RNTI = 112
 Handoff between cell 60 and cell 50

44
LTE location leaks and potential target device tracking
• According to 3GPP TS 36.300, 36.331, 36.211, 36.212, 36.213, 36.321
– C-RNTI is a unique identification used for identifying RRC Connection and scheduling which is dedicated to a particular
UE.
– After connection establishment or re-establishment the Temporary C-RNTI (as explained above) is promoted to C-RNTI.
– During Handovers within E-UTRA or from other RAT to E-UTRA, C-RNTI is explicitly provided by the eNB in
MobilityControlInfo container with IE newUE-Identity.

• No specific guidelines on how often to refresh the RNTI and how to assign it
– In my passive analysis I have seen RNTIs unchanged for long periods of time
– Often RNTI_new_user = RNTI_assigned_last + 1

45
Challenges and solutions
• Potential solutions
– Refresh the RNTI each time the UE goes from idle to connected
– Randomize RNTI
– Analyze the necessity of explicitly indicating the RNTI in the handover message

• If RNTI is not refreshed rather frequently

– MIT+Bell Labs work - LTE Radio Analytics Made Easy and Accessible (SigComm’14)
– Track a device and map measurements to it based on RNTI (paper’s section 8.7)
– When RNTI changes, PHY layer measurements still allow to map it to a given UE (SINR, RSSI, etc)
• MIMO measurements and metrics

• Recent discussion with GSMA

– The RRC Connection Reconfiguration message should be sent encrypted – This would make tracking more difficult
– But one could monitor traffic from adjacent cells and wait to see new RNTI with similar RF/traffic signature
– Ongoing discussions to address these potential issues

46
 Some final thoughts…

48
LTE security and protocol exploits
• Mobile security research very active since ~2009
• Most cool mobile security research exclusively on GSM (until now)
– GSM location leaks (NDSS’12)
– Wideband GSM sniffing (Nohl and Munaut – 27C3)
– Hijacking mobile connections (Blackhat Europe’09)
– Carmen Sandiego project (Blackhat’10)
– GSM RACH flooding (Spaar – DeepSec’09)
– …
• Recent availability of open source LTE implementations
– I expect a surge in LTE-focused security research
– Very interesting PhD topic

• The more research in the area, the more secure networks will be

• I am actively advocating for specific protocol security focus in 5G and next-gen standards
49
Thanks!

Q&A

http://www.ee.columbia.edu/~roger/ ---- @rgoestotheshows

Huge THANK YOU to Sanjole for providing the captures used in this presentation.
50 http://www.sanjole.com/