Security Of Wireless Networks: How Low-Layers Security Can Help

1. Security Of Wireless Networks:  How Low-Layers Security Can Help Authors: Date: 2008-11-11 Alex Reznik, InterDigital

2. Abstract Following up on the authors’ earlier presentation to the WNG (IEEE 802.11-08/0973r0), this contribution reviews the case presented for the need for low-layer security. We then provide additional detail on how low-layer security may be implemented and the potential impact on existing 802.11 specifications and implementations. Alex Reznik, InterDigital

3. Automation WLAN The NET Local WLAN WLANs in the Emerging World • Usage • Data (IP) access to a broadband network • Streaming applications over IP (e.g. VoIP, Video) • Interactive local application (distributed gaming) • Localized Mesh network with limited/no extra-net access • Machine-to-Machine communication • Home and Small Enterprise Automation • SensorNet-to-DataNet interaction • Distributed Computation • Security requirements • Secure transmitted data • Access control based on • Pre-shared keys (e.g. WPA) • ID (user and device) based access • Enforce network usage policies • Mesh security without a third-party certificate authority • Ensure network availability • Location-based requirements Alex Reznik, InterDigital

4. Moving Forward On Security • Currently existing security components provide: • End-to-end data protection • Limited or non-existing WLAN security components: • Device security • Device/host identity capability • Combined with trusted computing concepts • Location privacy assurance techniques • Local network security • Protect availability/access/usage of the physical medium • Enable location-based policies • These impact all layers of the architecture • Commonly used higher-layer protocols need to be enhanced with existing and new ones (e.g. IETF’s HIP, PBS, etc.) • Security Manager is needed for integration of higher-layer security policies and requirements with transmission medium capabilities • MAC support is required to make existing protocols attack-proof (e.g. CSMA), provide support for higher layer protocols (e.g. PBS, 802.1X-type authentication, etc.) and interface with PHY • PHY is required to monitor the medium, implement protocols as required by MAC • At all levels, services provided by secure platforms (e.g. via TPM) may be required • In this talk: • Concentrate on protection/access/usage of the physical medium • This is directly in scope for 802.11: as it inherently relies on PHY and MAC mechanisms Alex Reznik, InterDigital

5. High-Level Threat Analysis for 802.11 Security Attribute Threat These vulnerabilities are broad when viewed from a network perspective and are subject to a broad number of different attacks. These are identified (and in part addressed by 802.1), but not by 802.11 Alex Reznik, InterDigital

6. Addressing the threats in a wireless setting • Challenges: • Confidentiality/authentication techniques cannot address DoS attacks aimed at the network itself • 802.1AE techniques need to be examined, but are insufficient in a wireless context • Smart Jamming can masquerade as generic interference • Not addressed by 802.1AE as the problem is absent in wired systems • Terminal location is a priori uncertain • No connection to a port that signals can be traced back to • Alternate “secure key source” for confidentiality/authentication of pre-802.11i messages is desirable • 802.1AE calls for cipher suites which, in principle, do not need to rely on 802.1X • None have been proposed to date • A potential approach for wireless systems • Use the richness of the wireless access medium • Enhance PHY/MAC based security tied to the wireless medium • Enhance existing security mechanism • This is commonly called “PHY Layer Security” Alex Reznik, InterDigital

7. PHY-Layer Security • “PHY-Layer” security may: • Exploit the physical properties of the wireless channel • Typically have PHY and the MAC aspects • Be complementary to other modern techniques for securing wireless nets • PHY-Layer Security may use the physical link as a non-repudiatable, shared, secret resource to: • Provide always-on, link-specific message stream authentication • Classify interference appropriately (malicious/benign) and apply appropriate counter-measures • Distinguish between co-located and distinctly located terminals • Derive and update secret keys based on link properties and without the need to use (and expose) authentication credentials • PHY-Layer Security may further use link and terminal capabilities to: • Null out an identified rogue terminal • Switch away from channels under attack • Proactively hop channels to confuse attacker Alex Reznik, InterDigital

8. PHY-Layer Security - Examples • Example Attack 1: A Low-Power Jammer • Attacking terminal reverses the CSMA process – transmits whenever it detects energy on the channel • Legitimate terminals forced into even increasing back-off • Average power required for the attack is low • Initial power consumption is relative high • After a fairly short period, most terminals are in a long back-off state • The CSMA protocol assists in the attack • Example Attack 2: Sybil Attack • A terminal uses multiple MAC addresses: • Increases it share of bandwidth utilization • Attempts a service specific DoS (e.g. multiple authentication/association requests) Alex Reznik, InterDigital

9. Example Attack 1: A Low-Power Jammer • Potential Detection and Mitigation: • Detection: • Continual power measurement during channel clear state and burst reception reveal that burst should be successfully received • When this is violated sufficiently often (i.e. we observe statistically “impossible” collision pattern), an attack is likely • If MIMO is present, direction of interference may be estimated and further found to be non-random • Mitigation: • Alert security policy manager to abnormal condition, its nature, and (if possible) approximate localization • Switch channels. If possible establish a dynamic channel hopping policy. • Change the back-off protocol to eliminate increasing the expected back-off time. This will make the attack costly (in terms of energy) and may drain the battery of a true low-power attacker • If MIMO present, null away interference source. Alex Reznik, InterDigital

10. Ex. 1: Implementation of Detection: in the PHY Alex Reznik, InterDigital

11. Ex. 1: Implementation of Detection: in the MAC • Comments: • 2 minor modifications in PHY • Reporting certain quantities (RSSI and Channel Clear) at all times as opposed to as a result of certain event • A new mechanism for providing this report to PHY is therefore required • No new PHY processing is needed • New MAC processing is added, existing functionality not affected Alex Reznik, InterDigital

12. Ex. 1: Implementation of Mitigation • Notes • Mitigation impacts only the Tx Control SEQR • This involves added functionality only • Currently: single Tx Policy (current CSMA/CD) • New: • a number of policies, of which existing is one • Tx Control/SEQR selects one based on request from Security Manager • No impact in the PHY • Expected for mitigation Alex Reznik, InterDigital

13. Example Attack 2: Sybil Attack • Potential Detection and Mitigation: • Detection • Using channel-based signatures establish the fact that multiple MAC addresses appear to be from same radio • Mitigation • Alert security policy manager to abnormal condition, its nature, and (if possible) approximate localization • If required by policy, establish that when treated as an aggregate these do not follow the proper protocol for a single terminal • De-associate all MAC addresses with suspect channel signature • If MIMO present, null away transmission from suspect location Alex Reznik, InterDigital

14. Ex. 2: Implementation of Detection • Notes • PHY Impact: • No new functionality • Reporting of CIR • MAC Impact • Increasing existing address matching functionality to check for signature/address consistency • Addition of “security alert/attribute” to each PDU • As before augmentation, not modification of existing functionality, Alex Reznik, InterDigital

15. Ex. 2: Implementation of Mitigation • Notes • Primary mitigation strategies involve management-level activities • Additional management procedures added to address security needs • PHY-based (MIMO) countermeasures require minor modifications to PHY beamformer Alex Reznik, InterDigital

16. Observations • Detection • Requires Physical Measurements to be made available … • SW based “detection” algorithms determine alerts • Mitigation: • Requires MAC Control of Parameters of Existing Protocols …. • Enables MAC to encrypt control frames as follows… • New control/man. Frames to support all this??? Alex Reznik, InterDigital

17. Review of Previous Straw Poll Results • Do you believe that future 802.11 systems/applications will require security beyond what 802.11 currently has? • Yes \ No \ Need more info \ Don’t care: 21 \ 11 \ 2 \ 1 • Do you believe that protecting against DOS attacks requires security beyond what 802.11 currently has? • Yes \ No \ Need more info \ Don’t care: 8 \ 2 \ 18 \ 0 • Do you believe that protecting against False Identity attacks requires security beyond what 802.11 currently has? • Yes \ No \ Need more info \ Don’t care: 6 \ 3 \ 22 \ 0 • Should 802.11 start a study group to address physical layer security? • Yes \ No \ Need more info \ Don’t care: 2 \ 15 \ 17 \ 0 Alex Reznik, InterDigital

18. Straw Poll 1 • Is detection of attacks like Example 1 and Example 2 desirable for 802.11? • Yes • No • Need more info • Don’t care Alex Reznik, InterDigital

19. Straw Poll 2 • Is mitigation of attacks like Example 1 and Example 2 desirable for 802.11? • Yes • No • Need more info • Don’t care Alex Reznik, InterDigital