230 likes | 374 Views
Antenna-Transceiver Integration. Meeting selectivity specs is one of the big challenges for this architecture Our approach: RF multiplexer optimized to antenna impedance with external noise dominance constraint No antenna tuning! Simultaneous access to multiple bands .
E N D
Antenna-Transceiver Integration • Meeting selectivity specs is one of the big challenges for this architecture • Our approach: RF multiplexer optimized to antenna impedance with external noise dominance constraint • No antenna tuning! • Simultaneous access to • multiple bands “Transducer power gain” for RF multiplexer optimized for a 20 cm long monopole antenna “External noise dominance” in VHF-High and 220 MHz bands RF Front End (RFFE) Board S.M. Shajedul Hasan and S.W. Ellingson, “Multiband Antenna-Receiver Integration using an RF Multiplexer with Sensitivity-Constrained Design,” IEEE 2008 Int'l Symp. Ant. & Prop.
Direct Conversion Transceiver Sections (2nd Gen.) These two boards stack vertically with the RFFE board using MMCX connectors (no RF cables) ADC / DAC / LO Synthesizer Board (2nd Gen.) ADC/DAC: 130 mA @ 9V, running 4 MSPS < 50 cm2 to implement on a 4-layer PCB ADC ~ $21 (1k), DAC ~ $10 (1k) 4-Band Transceiver Board (2nd Gen.) 40 mA (RX) + 40-90 mA (TX) + 80 mA/DDS @ 9V < 25 cm2 to implement on a 4-layer PCB About $100 in parts to implement, excluding PCB.
Baseband Design Actually Used • Prototype currently implemented on • an Altera Stratix II FPGA (EP2S60). • Massive overkill (60k LEs) but board • familiar and readily available. • Implemented directly via FPGA: • Analog FM waveform • Synthesis of 4 MHz ADC/DAC clocks • Interfaces to codec, ethernet, and display controller chips • In our project, it appears that we will not need to implement a soft-core processor: We are currently 100% Verilog HDL. • Target FPGA is Altera Cyclone III; 25k LEs, about $50 (chip). Power consumption is variable and hard to predict, but state-of-the-art power management features are available.
Thanks! Acknowledgements: Motorola: G. Cafaro,B. Stengle, N. Correal Mahmud Harun (student) Rithirong Thandee (student) Web Sites: http://www.ece.vt.edu/swe/chamrad/ http://www.ece.vt.edu/swe/ http://wireless.vt.edu/ U.S. Dept. of Justice National Institute of Justice Grant 2005-IJ-CX-K018
SDR/CR Security Jung-Min “Jerry” Park and Tim Newman
Security Issues in Cognitive Radio Networks Jung-Min “Jerry” Park
SDR/CR Software Tampering • Adversary can alter radio operating characteristics by modifying SW • Software critical to radio operating characteristics • Policies (policy database) • System strategy reasoner • Policy reasoner XG radio architecture [3]. [3] D. Wilkins et al., “Policy-based cognitive radios,” IEEE Wireless Comm., Aug. 2007.
Our Approach for CR Software Tamper Resistance Applying tamper resistance to unprotected assembly code. Contribution of each operation on runtime overhead.
Testing & Implementation of Our Scheme Tamper resistance primitives embedded into assembly code GNU Radio USRP board Tamper resistant executable code
Policy-related Security Issues • Policy enforcement • Need to ensure that SDR’s configuration conforms w/ regulatory and system policies • Need a “Policy Enforcer (PE)” • Erroneous radio operation due to policy conflicts • CRs operate in the presence of multiple policies from multiple stakeholders • When multiple policies are activated, CR needs to resolve conflicts among policies [6] F. Perich et al., “Policy-based spectrum access control for dynamic spectrum access network radios,” Web Semantics Sci Serv Agents World Wide Web, 2008. [7] P. Flanigan et al., “Dynamic policy enforcement for software defined radio,” 38th Annual Simulation Symposium, 2005. Policy enforcement in XG radio [6]
Need for CR Policy Analysis • Motivation for policy analysis • Need to identify policy conflicts • Policy analysis results can be used to aid • Policy Reasoner’s computation of opportunity constraints • Creation of meta-policies for prioritization of policies • Shared Spectrum Company’s policy conflict resolution scheme consists of a combination of a default rule and a prioritization schemata Spectrum access opportunity discovery in XG radio [6]
Our Approach for CR Policy Analysis A graph-theoretic approach for policy analysis [8] G. Denker et al., “A policy engine for spectrum sharing,” IEEE DySpan, Apr. 2007. SRI International’s policy reasoning procedure [8]
Cognitive Radio Security Issues (Overview) Timothy R. Newman, Ph.D.
Cognitive Radio Introduction • Cognitive Radios (CR) and CR Networks offer amazing promise • Intelligent Radios • Self Adapting • Improve Overall Communications • Primary Focus of research has not been security • DSA • Cognitive Engines (GA/CBR/Rule-based) • New Technology means new Threats • Putting AI in charge of radio operation causes potential problems
Cognitive Radio Introduction • Multiple Classes of new CR specific attacks • Sensory Manipulation • Belief Manipulation • Cognitive Radio Viruses • Common goal of creating sub optimal communication or malicious communication • CRs need common sense in order to overcome these attacks
Policy Radio Threats • Policy Radios • No learning involved • Basically just a validation and recommendation engine • Shared Spectrum Radios are DSA Policy Radios • Primary concern is sensor spoofing or “sensory manipulation attacks”. • Rely on knowledge of internal logic • What types of inputs are being used • How these input statistics are calculated • How they affect the transmission parameters – (e.g. the fitness function)
Learning Radio Threats • Vulnerable to the same threats as Policy radios with an added twist. • Learning radios make decisions based off previous experiences • Threat damage can be long term • Example: Jam the communication of a learning radio whenever it uses a fast modulation rate. • This TEACHES the radio to that fast modulations produce an extremely high BER. Forcing it to use lower modulation rates. • Learning radio may store this in memory after X number of attacks causing long term damage. • Cognitive Radio Networks can proliferate malicious actions. • The state of Radio 1 can affect the state of Radios 2,3, and 4. • Cognitive Radio Network virus!
Objective Function Attacks • On path attackers can determine key characteristics • Symbol Rate • Modulation • Manipulation the beliefs of the CR using environment parameters • Example: Jam channel when using High Security • Result: Fitness function is always higher when security is low • Overall objective is to make the CR believe specific options are not optimal • Attacks restricted to CRs that use online learning f = w1P + w2R + w3S
Individual CR Attack Mitigation • Need to instill some “common sense” into the radios. • Robust Sensor Inputs • Ideal CR can always tell the difference between interference and noise • Distinguish between natural and man-made RF • Robust Data Fusion Techniques for CR Networks • Distributed Environment technique used to improve performance • Decision Fusion • Bayesian Detection • Neyman-Pearson Tests R. Chen, J.-M. Park, Y. T. Hou, and J. H. Reed, "Toward secure distributed spectrum sensing in cognitive radio networks," IEEE Communications Magazine Special Issue on Cognitive Radio Communications, Apr. 2008.
Individual CR Attack Mitigation • Individual policies should be developed with care • Similar to writing robust code. Sure strcpy() works but its extremely exploitable if not used properly. • Formal state-space validation can be done to ensure bad states can not exist. • PUE attacks require a different more technical approach • Develop better sensing algorithms! • Energy detection is extremely simple but should practically never be used by itself outside of a lab. • Cyclostationary features • 10-20 Time domain characteristics
Individual CR Attack Mitigation • A priori information can be used to supplement sensing techniques • Geolocation – Validating physical location using SNR • FCC white space decision will require this most likely • Learning attacks can be mitigated through constant feedback. • Receiver is constantly providing QoS feedback so learned actions do not get poisoned. • Not full proof but requires attacker to be actively “teaching” in order to keep the learned actions invalid. • Develop relationships or objective functions that demonstrate orthogonality between objectives that aren’t related. • Example: Throughput should not affect security level • In no situation should we adapt the security level simply because the throughput is lowered. • This results in a more complex fitness function but removes a large portion of the ability for attackers to manipulate unrelated objectives.