Summary of Projects - Virginia Tech of Projects May 2012 Wireless @ Virginia Tech Virginia Polytechnic Institute and State University 432 Durham Hall Mail Code 0350 Blacksburg, VA

Summary of Projects

May 2012

Wireless @ Virginia TechVirginia Polytechnic Institute and State University

432 Durham HallMail Code 0350

Blacksburg, VA 24061Tel: 540-231-2958

http://www.wireless.vt.edu

The Android Tactical Application Analysis & Knowledge Cloud, Jules White

Goals

ApplicationsHighlights

- Highly Configurable/Connected Smartphone Emulation- Large-scale Multi-application Test Scenarios- Real-world experimentation and data integration

- Launch 1000s of emulators simultaneously - Identify 1000s of valid configurations- Run testing code across range of emulators, and collect output log, performance, and visual data

- Enable production-scale testing of smartphone-dependent systems using actual application code- Enable agent-based simulations in mobility and social models - Measure mobile malware spread on modern Android devices

AUSTIN: An Initiative to Assure Software Radios have Trusted Interactions

Jeff Reed, Tom Hou, Jung-Min Park, Cameron Patterson

Objective /Goals

Impact and ScopeHighlights

• Examine the theoretical underpinnings related to distributed system regulation for software radios

• Develop an architecture that includes trusted components and a security management plane for enhanced regulation

• Research onboard defense mechanisms that involve hardware and software-based security

• Study algorithms that conduct policy regulation, anomaly detection/punishment, and secure accounting of resources

• Holistic approach for regulating software radios

• Execution Integrity of Critical Software Modules in CRs

• Using Power Fingerprinting to detect illegal execution

• Policy Reasoning Algorithms and Policy Reasoners

• Policy reasoning algorithms and software engines for real-time policy interpretation, process, and enforcement

• Network Resource Optimization for CR Networks

• Unifying analytical framework for the multi-dimensional cross-layer design space of CR for tractability

• Configurable Hardware-Assisted Rule Enforcement Controller for CR reconfigurable platforms

• Automatically integrates PHY- and MAC-level policy enforcement into CR reconfigurable hardware blocks

Period of performance: Sep 2009 – Aug 2011Sponsor: National Science Foundation

• Ensure the trustworthy operation of software radios

• Support the next generation of wireless technology

• Holistic system view

• Deeper understanding of security for highly-programmable wireless devices

• Reduce interference risks and increase trust in CR Technology

• Diverse, multi-disciplinary team

• WINLAB at Rutgers University

• Wireless @ Virginia Tech

• University of Massachusetts

Ana

lysi

s

De

sign

Po

licy

rea

son

ing

a

nd e

nfo

rcem

ent

Inte

grity

Ass

essm

ent

Tru

stw

ort

hy

Con

figu

rab

le H

W

Trustworthy CR Operation

Coexistence Decision Making for Spectrum Sharing among Heterogeneous Wireless Systems.

Objective

Impact and ApplicationsHighlights

PI: Jung-Min “Jerry” Park

• Design and analysis of a coexistence decision making (CDM) algorithm for a system of coexistence mechanisms, such as an IEEE 802.19.1 compliant system that enables coexistence of dissimilar TV whitespace networks and devices.

• The algorithm should focus on ensuring fairness and satisfying coexisting networks demands in allocating spectrum to heterogeneous wireless networks.

• Use a set of constraints to formulate the CDM problem as a multi objective combinatorial optimization problem.

o Each of the constraints address a challenge or requirement of a coexistence system for heterogeneous networks.

• Model the optimization problem as an energy minimization problem in a Boltzmann machine.

• Propose an algorithm to find a near-optimal feasible solution in the Boltzmann machine.

• One of the first works to address the important problem of heterogeneous coexistence of secondary networks/devices.

• A coexistence decision-making algorithm that outperforms other CDM algorithms in terms of fairness, packet error rate and percentage of demand serviced.

A coexistence enabling system with a centralized topology.

NSF-REU: Cognitive CommunicationsTamal Bose, Tonya Smith-Jackson

Objective /Goals

Research Areas

Highlight

To gain hands-on experience in cognitive radios, wireless networking, and their applications;

To develop general research skills and teamwork skills;

To expose the students to the intellectual excitement in emerging wireless communications technology;

To promote creative and independent thinking, a sense of confidence, team spirit, and an appreciation for the potential of interdisciplinary collaboration in creating new knowledge;

To excite the students to pursue graduate studies.

An opportunity for undergraduate students from around the country to participate in the ongoing research activities of Wireless@VT.

Theory and application of algorithms and artificial intelligence for wireless communications;

Signal Processing for software defined radio and cognitive radio communications;

Software framework (OSSIE/GNUradio) and its applications for software defined radio and cognitive radio communications;

Algorithms implementation using a cognitive radio network testbed(VT-CORNET)

Applications of usability/human factors principles, and sociotechnical systems frameworks to cognitive engine and interface design, development, and evaluation.

Sponsor: NSF

Period of Performance: 2009-2015

Cognitive Jammer Detection and ClassificationPI: Jeff Reed

Objectives/Goals

ImpactsHighlights

• Develop an engine to detect the presence of a cognitive jammer, as well as traditional jammers• Once detected, classify the jammer in terms of behavior, signal characteristics, spectral characteristics, and any other key features that can be extracted• Acquire information about the jammer that will aid in dynamic routing

Period of performance:January 2012 – January 2014

Sponsor:Office of Naval Research

• Developed a set of detailed jammer models, ranging from simple to extremely sophisticated• Created high-level engine that classifies type of jammer, given the jammer and channel activity• Analyzed different methods of determining jammer’s presence when it is covered by a friendly radio’s signal

• Enables data delivery in near real time under extremely harsh networking conditions that include RF interferences created by the intentional and cognitive jammer• Supports the next generation of electronic warfare support

Cognitive Radio Network Testbed (CORNET)PIs: Jeff Reed and Tamal Bose

Objectives/Goals

ImpactsHighlights

• Develop and build an open heterogeneous cognitive radio network testbed for the development, testing, and evaluation of cognitive engine techniques and cognitive radio network applications.

The ICTAS building with the testbed nodes in red

Period of performance:September 2009 – Current

Sponsor:Office of Naval Research and ICTAS

• 48 nodes in a modern engineering building• Node hardware:

• Dedicated GPP server for each node• USRP 2• Custom RF DB (100MHz- 4GHz)

• Customizable PHY and MAC layer • Remotely accessible• Free, Open Source Software

• GnuRadio• OSSIE• CROSS• LiquidDSP

• Unique CR prototyping platform• Wireless application testing in a live spectrum environment• Broad FCC experimental license allows complete utilization of USRP tunable range• Sponsored research and remote education• Indoor node distribution provides diverse channel conditions• Valuable learning environment using real radios

NSF-CCLI: Cognitive Radio Vertical IntegrationTamal Bose, Tonya Smith-Jackson, Carl Dietrich

Objective /Goals

Impacts Highlights

v

Introduce Cognitive/Software Radio concepts to Undergraduates. Move current Graduate level SDR course into the undergraduate curriculum. Vertically integrate CR/SDR concepts into several undergraduate courses, starting Freshman year. Adapt and expand current curriculum to better prepare graduates for modern communications topics. Develop lab activities and workshops to provide hands-on learning.

Development of undergraduate lab activities at VT and NSU Digital Communications: SDR Using OSSIE

Analog AM/FM analysis Digital PSK/QAM/ASK analysis

Engineering Exploration: Communications Intro using Matlab

Intro to digital modulation and BER Pulse shaping and channel filters

Trial labs run during Fall semester 2010 Vertical integration self assessment

Preliminary integration of lab activities into the undergraduate curriculum provides a basis for the continued vertical integration of advanced communications topics into the undergraduate curriculum.

Cooperation with NSU facilitates a multi-institutional plan for implementing and analyzing amendments to the curriculum across a diverse cross section of students.

Sponsor: NSF

Period of Performance:2009-2012

Collaborative Localization Enhancement to GPSZachary Biskaduros & R. Michael Buehrer

Objective /Goals

Impacts Highlights

• Explore collaborative techniques in positioning location.• Create a MATLAB simulation to mimic a GPS receiver

with all 31 operational satellites allowing us to alter the environment of the user and examine the errors that occur due to noise and limited satellite availability.

• Introduce a new solution that has the ability to improve the accuracy of a GPS receiver when in a challenging environment by introducing additional user(s) to the GPS solution with the knowledge (or estimate) of the distances between the users.

• The new solution allows our GPS receivers to collaborate with each other and solve all positions simultaneously.

Oak Ridge National Laboratory

2012

• Simulation results show improvement in both the accuracy of the position estimate and availability

• The new collaborative solution using inter-receiver distances along with the standard pseudoranges to cooperatively determine all receiver locations simultaneously

Collaborative Research: Fundamental Research on Adaptive Wireless Video Systems

PIs: Haris Volos and Tamal Bose

Objectives/Goals

ImpactsHighlights

• Wireless distributed video processing and delivery of streaming concurrent variable-bit-rate (VBR) videos• Proof of concept application using VT’s 48 node cognitive radio wireless testbed (CORNET•Investigate the effect of different metrics such as frame rate, resolution, and power consumption on the performance of the application

Period of performance:September 2011 – July 2012

Sponsor:NSF

• Collaborative project with the WICAT center partners •A prime example of a wireless distributed computing (WDC) application• The video processing algorithm can be applied to consumer (cellular and home networks) and military applications.

• Our results show that a distributed algorithm that takes in account the channel heterogeneity is the key in order to build power efficient and robust WDC networks.• Predictable, variable quality of service is of paramount importance for wireless applications• Leveraged VT-CORNET

Computational Social Science : Modeling Human BehaviorC. Kuhlman, M. Marathe, H. Mortveit, S. Swarup, G. Tuli, A. Vullikanti, S. Ravi, D. Rosenkrantz

Virginia Bio-informatics Institute and Dept. Computer Science, University at Albany NY Dept. Computer Science

Objectives /Goals

Impacts Highlights

Sponsors: DTRA, NSF, DARPA, IARPA

• InterSim: An HPC-based modeling framework (based on Graph Dynamical Systems theory) to study simultaneously (almost) any collection of social processes.

• Algorithms for both fostering and inhibiting complex and generalized contagions.

• New bi-threshold model and cooperative action models.• Phase-space properties of bi-threshold systems proved,

which shows the permissible regimes of long-term behavior of these systems.

• Methodology applied to youth smoking initiation and cessation using Ad-Health database.

Challenges:• Myriad application domains

(e.g., viral marketing, social unrest, youth smoking).

• Wide ranging human behaviors in one domain.

• Myriad interaction mechanisms.

• Scalability.

InterSim can implement (almost) all behavioral models we’ve found to date in literature, giving us tremendous expressive power in simulations.Methods for blocking diffusion are far superior to

previously considered best performing algorithms. Can be used to block harmful rumors, etc.Our published ideas for fostering diffusion, based on k-

cores, have been subsequently verified to occur in real systems [Nature Scientific Reports 2011]. Proofs of phase space behaviors save great deal of

modeling time by guiding our simulations.

Results published in: PKDD 2010, Automata 2011, CAS-AAAI 2011, SBP 2011, SNA KDD 2011, WSC 2011 (x2), WebSci 2012.

Many models of social behavior developed on small populations. With social media, models will evolve and need to be applied to large populations. Results needed quickly in dynamic environments.

Egypt, Arab Spring, 2011

Reuters

gossip

peer influence, youth smoking

Modeling: (i) simulation of million-plus agent systems over wide ranging application spaces with multiple contagions; (ii) develop new behavior models. Algorithms: develop methods for controlling contagion

processes, with accompanying theoretical justifications. Theory: provable behavior of different systems.

Cross Layer Control Mechanisms Inspired by Game TheoryWilliam Rogers, Allen B. MacKenzie

Objective /Goals

Impacts Highlights

• Develop cross-layer game-theory-inspired, near-real-time control mechanisms

• Assess convergence rate under steady state conditions

• Assess network overhead induced by exchanged messages

• Create a detailed plan for the design and development of a prototype system

• Developed a two layer game model to study control mechanisms in wireless networks

• Created a high layer Topology Control game with mission dependent objectives that decides which links are kept active

• Created a low layer Link Optimization game that optimizes the links chosen in the topology control game

This material is based upon work supported by US Army CERDEC and the National Science Foundation under Grant No. 0809036.

• Real time control mechanism that makes a wireless network submissive to the mission

• Plan for game model to be tested experimentally in VT’s CORNET USRP testbed

A randomly plotted topology of 30 nodes in a 10x10 grid that was used in the Topology Control game.

Cross-Layer Optimization for Video Transport in Wireless Ad Hoc Networks

Investigators: Thomas Hou, Hanif D. Sherali

Objectives

• Support video communications over dynamic ad hoc networks

• Mitigate the impact of network dynamics (topology change, link/node failure) on video quality

• Optimize video quality under network resource constraints

Impacts

• Numerous applications for public safety, military, and general public

• Serves a critical need in ad hoc networking community

• New educational materials for classroom and research opportunities for students

Highlights

• Exploit multiple description video and multipath routing

• Cross-layer optimization with application layer performance objective

• Design of distributed algorithms and protocols

• Testbed prototyping and validation

Graphic or Photo

Period of Performance: 8/2009 – 8/2013Sponsor: NSF

CTrackAlan Baines, Jules White, and Brian Dougherty

Objective / Goals

Highlights

Graphic or Photo

Period of perfsponsor

Using smart phone to automate the process of monitoring construction site workers by incorporating GPS and accelerometer data collected.

Organize the data in a format that Architecture / Engineering / Construction field personnel can use to increase the efficiency of construction sites.

Using Fast Fourier Transforms of accelerometer data and BayesClassifiers including GPS data to determine the type of activity that construction workers are currently performing, as means to get past noisy and unreliable sensor data.

Research would allowConstruction Engineers to testand experiment with different organization and planning techniques quickly and cheaply while also being consistent.

Disaggregated, Spatio-temporal Modeling of Spectrum demandMadhav Marathe, Achla Marathe, Anil Vullikanti,

VBI, CS and Ag and Applied Economics

Objective /Goals

Impacts Highlights

Sponsors: NSF, DTRA

Significant white spaces in many spectrum bands

Cognitive networks for dynamic spectrum access

Auctioning of spectrum bands in primary and secondary markets

Spatio-temporal model of spectrum demand Depends on underlying cellular

infrastructure, social network, Network planning: model effect of

behavioral changes on spectrum demand Understand the impact of different kinds

of market mechanisms

Spectrum demand in Portland, OR at 1:00 pm

Activity based framework for modeling spectrum demand in urban regions Integrates a number

of public and commercial data sets

Agent based framework for studying different market mechanisms

Can study impact of behavioral changes and homophilies in call patterns

Speculators and secondary market has significant impact on market efficiency

Truthful auction mechanisms such as Ausubel’s ascending auction can be applied for efficient market clearing.

Published in [Beckman et al., IEEE DySPAN 2010], [Kim et al., IEEE DySPAN 2011], [Barrett et al., CRIS 2010]

Distributed Radio Environment MapsAbid Ullah, Joseph D. Gaeddert, etc.

Objective/Goals• Store/retrieve geospatial information

about wireless environment

• Integrate distributed spectrum sensing with cognitive radio collaboration

• Develop algorithms for efficient exchange of spectrum information

• Verify simulation environment with results from measurement campaign

Impacts • Coexistence between radar and

communications signals

• Open-source radio environment software for researchers

• Extensible to unknown/undeveloped protocols

• Does not require special hardware (software radio not a necessity)

Highlights• Non-centralized database

• Enhanced security, faster convergence over centralized design

• Protocol agnostic, extensible

• Implemented with a relational database management system (SQL)

• Open-source software

• Demonstrated on CORNET (cognitive radio testbed)

Economic Models for Collaborative Access Network ProvisioningUS-Finland Collaboration

Dr. Allen B. Mackenzie, Dr. Luiz A. DaSilva

Objective /Goals

Impacts & Applications

• Solution tailored to a real world hybrid network consisting of large scale cellular network in cooperation with a small-scale WLAN.

• Support efforts towards network interoperability and cooperation.

• Work to be disseminated broadly through scientific conference and journal papers.

Highlights• Developed an initial game theoretic model of incentives

between small (WiFi) and large (cellular) network operators.

• Will create a model of a complex, heterogeneous network in a wireless testbed setting.

• International collaboration (US and FIN) provides unique experience of interdisciplinary working with enhancing inter-cultural understanding.

• Effort to educate students and practitioners about the project through short courses, tutorials, and seminars.

• Game theoretic models to assess incentives and pricing mechanism for collaborative network consisting of large mobile operators, and small (campus/home) operators.

• Concurrent routing and forwarding algorithms for multi-hop/operator/technology network, quantifying capacity extension due to collaboration.

• Threat models and security approach for collaborative networks.

• Testbed demonstrations to prove concepts and techniques.

This material is based upon work supported by the National Science Foundation under Grant No. 1147790.

Efficient Jammers Using a Cognitive Radio NetworkTamal Bose and Jeffrey H. Reed

Objective /Goals

ResultsHighlights

Proposed Cognitive Jammer

Sponsored by: Center for Advanced Engineering and Research

Automatic modulation classifier (AMC) that is robust to multipath effects was developed as a component of the CJ unit.

CJ unit was implemented using USRP and GNU Radio.

Covert Jamming was demonstrated using the proposed testbed

• Malicious users can easily identify the presence of conventional jamming signals and adapt their operations accordingly.

• The goal of this project is to develop Cognitive Jammers (CJ) that overcome the disadvantages of conventional jammers by covertly jamming the malicious users.

• The CJ must also offer least disturbance to friendly users in the frequency band.

The proposed CJ generates a covert jamming signal by taking into consideration the following

•Modulation type used by the malicious users•Type of users operating in the frequency

band•Channel conditions•Energy in the band•Past experience

A three radio cognitive jamming testbed was proposed to test the proposed CJ unit.

Exploring Performance Limits of Multi-hop MIMO Networks via Tractable Models and Optimization

Investigators: Thomas Hou, Scott F. Midkiff

Objective and Approach• Observation: Rapid advances of MIMO

research at the physical layer, but a stagnation of research progress at the network layer

• Our goal: Exploring fundamental understanding of multi-hop MIMO networks

• Approach: Build new tractable and accurate models that are amenable for network level optimization

Impacts• Remove the current barrier facing multi-

hop MIMO network research

• Offer new network-level understanding of MIMO networks

• Bridging the gap between wireless communications research and networking research

• Develop new cross-disciplinary course materials for wireless networking

Highlights

• Develop new tractable and accurate models for MIMO links

• Explore performance limits and optimization for multi-hop MIMO networks

• Design distributed optimization algorithms

Graphic or Photo

Period of Performance: 7/2011 – 6/2014Sponsor: NSF

The Flexible Internetwork Stack (FINS) FrameworkA. B. MacKenzie and L. A. DaSilva

Objectives

Development of an open source reconfigurable protocol stack for experimental research on wireless networks.

FINS will provide modules: To build conventional as well as innovative network

stacks using existing or new layers, protocols, and algorithms.

To monitor, visualize and log the behavior of all the stack components.

To change the behavior of the networking stack by modifying the behavior of individual modules or reconfiguring the stack, while it is in use (On The Fly Re-configurability).

Why FINS ? The growing doubt about the fidelity of simulations as the

main technique for evaluation of newly proposed network protocols and algorithms.

The need to overcome technical and logistical challenges in experimental network research.

The lack of well documented, continuously supported as well as easy to use educational experimental networking tools to help the researchers to access the implementation of the network stack.

The lack of enabling technologies for implementing of cyber physical systems prototypes.

Use Cases Prototyping experimental cross-layer solutions.

Prototyping delay and disruption-tolerant networking protocols such as bundle and routing protocols.

Implementing experiments for reconfigurable cognitive networks as well as cognitive nodes.

Implementing wireless ad-hoc network experiments like MANIAC competition.

Experimental context-aware solutions (e.g., HTTP Adaptive Video Streaming Applications).

FINS vs Conventional TCP/IP Stack

Current Status For latest updates check the FINS Framework official page www.finsframework.org The latest available version is v0.9. Download as tarball file or using Git from https://github.com/mackenab/FINS-Framework.

This material is based upon work supported by the National Science Foundation under Grant No. 0916300.

Generalized Bayesian Position EstimationJavier Schloemann & Dr. R. Michael Buehrer

Objective /Goals

Current Research FocusHighlights

Research sponsored through the Bradley Foundation

Particle Filter-BasedCollaborative Localization & Sensor Fusion

Generalized Bayesian Position Estimation FrameworkSingle-Node Static Localization•Memoryless localization of a single node using only external sensor measurementsSingle-Node Mobile Localization•Filtered localization of a single node using past location, movement, and external measurement history along with present movement data and external sensor measurementsMulti-Node Static Collaborative Localization•Memoryless localization of multiple nodes using standard external sensor measurements along with relative measurements between cooperating and collaborating nodesMulti-Node Mobile Collaborative Localization•Filtered localization of multiple nodes utilizing all past and present information shared between cooperating and collaborating nodes

• Simulations• Individual monitoring system using disparate data

combined (or fused) using particle filters• Collaborative robot localization using particle filters

with and without anchors• Characterized the value of collaborative localization using

particle filters to approximate the a posteriori location beliefs at every robot

• Benefits in RMSE performance• Reduction in belief standard deviation• Noted introduction of universal localization bias

• Successfully utilized disparate data including RFID proximity sensors, optical data, and TDOA radio positioning

• Perform a mathematical characterization of estimation error using uncertain anchor and robot positions in collaborative localization

• Analyze the effect of geometry on estimation error

• Complete the characterization of the value of collaborative localization in environments with and without anchors

• Extend work to focus on the tougher indoor localization scenario using NLOS signal propagation

Intelligence Community Center of Academic Excellence (ICCAE)Reed, Bose, Amanna

Objective /Goals

Impacts Highlights

Period of Performance 2009-2014Sponsor: ODNI

• Develop critical thinking and core competencies in new graduates• Increase awareness of the missions of the IC and National Security• Sponsor study abroad to increase cultural awareness• Develop new curriculum• Host symposiums on IC / National Security Topics• Encourage and assist with internship applications• Sponsor pre-collegiate outreach• Engage with IC for guest speakers and student mentoring

I just wanted to give you a big ‘thank you’ for all that you’ve done with the IC-CAE program. I know that you’ve put a lot of time and effort into helping us out as students and as young professionals. I can’t explain how helpful this program has been and how many doors it truly has opened for me.” – Lucas W

“Again, thank you for all your support and guidance over the pastyear! I feel like IC CAE assisted me so much in my job search process and has been such a fantastic program to be a part of, I've enjoyed every minute. Please let me know if I can assist in any other way or if you have any advice for me as I continue in my career after graduation.” –Kammy M.

21 students sent on study abroad since 2009

104 applications to IC agencies for 2011

Over 100 office hours for 2011

9 IC Guest Speakers for 2011

6 COE to Govt. agencies

4 new courses, 1 certificate, 1 minor, in proposal

STEM outreach to Middle school

Infrastructure InterdependencyMadhav Marathe, Achla Marathe, Anil Vullikanti,

VBI, CS and Ag and Applied Economics

Objective /Goals

Impacts Highlights

Sponsors: NSF, DTRA

Need tools to study interdependencies in the networked infrastructures.

End to end simulation system that can not only model and simulate multiple infrastructures but also capture the points of interdependencies among them.

Systematic study of human-initiated disruption, resulting in cascading failures in infrastructures:Three networks are considered:

Transportation networks, social phone call networks and cellular networksNetwork planning: model effect of changes

in call patterns during evacuation and its impact on transportation and spectrum demand

Activity based framework for modeling traffic congestion, revised cell usage patterns in face of evacuation caused by a chemical plume in downtown Portland Oregon.

Study how altered human behavior at the time to crisis causes altered activities, travel, route planning and calling behavior.

Study the cascading effect of congestion in transportation network on overloading of base stations and altered spectrum demand.

Published in [Beckman et al., IEEE DySPAN 2010], [Kim et al., IEEE DySPAN 2011], [Barrett et al., CRIS 2010]

Low-cost Cognitive Radio Technology for Enhanced Communication and Situational Awareness for Networks of Small UAVs

Charles W. Bostian

Objective /Goals

ApplicationsHighlights

This material is based on research sponsored by Air Force Research Laboratory under Agreement #FA8750-09-2-0128. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of Air Force ResearchLaboratory or the U.S. Government.

• Implement CSERE cognitive engine (CE) on BeagleBoard single-board computer

• Provide USAFRL Rome with prototype flight model CE and cognitive radio CR called UAV CSERE, suitable for deployment on USAFRL UAVs

• Cooperate with USAFRL to develop and conduct live-flight UAV CR experiments for UAV CSERE

CSERE: CWT’s self-modifiable, modular cognitive engine • Mission resource

conservation: UAVsshare responsibility for data backhaul

• Link reliability: UAVs with poor link performance hand off to with better link performance

• Mission Optimization: UAVs optimize link andposition to ensure optimal mission performance

Period of Performance:10/20/2010 – 9/30/2012

Mobile Cognitive Radio TestbedPI: Tamal Bose

Objectives/Goals

ImpactsHighlights

• Study research issues pertaining to the mobility of nodes •Test developed algorithms, protocols, and applications in a realistic mobile wireless environment • Reconfigurable mobile network• Connect with the network testbed of fixed CR nodes to have larger scale heterogeneous testbed

Example scenarios that can be evaluated with the proposed mobile nodes

Period of performance:September 2010 – December 2011

Sponsor:DOD DURIP2

• Custom mobile node that includes:• A Small Laptop PC• Pico Expresscard (Virtex-5 FPGA)• Custom RF board (100Mhz to 4GHz)

• Will facilitate the evaluation of scenarios that involve mobile nodes• Will enhance the capabilities of the existing fixed node testbed, CORNET• Can be used both for educational and research purposes

Mobile Content Distribution in Vehicular Ad Hoc NetworksPI: Wenjing Lou

Objectives

• Application scenario: Mobile content distribution (MCD) in vehicular ad hoc networks (VANETs)

• Challenges and opportunities: high throughput, low latency, high mobility, common interests of receivers

• Main technique: Symbol level network coding (SLNC)

Highlights• Symbol level diversity leads to higher throughput

• Higher spatial reusability leads to higher throughput

Sponsored by National Science FoundationSep 2011 ~ Aug 2014

Impacts/Applications

• Theoretical models for SLNC-based network are developed in a flow network setting, which provide a base for capacity characterization of SLNC-based mobile networks.

• Distributed and localized algorithms and protocols for mobile networks using SLNC are developed to exploiting the advantages of SLNC fully in a MCD network.

Modeling and analysis of malware in 3G+ networksM. Marathe, A. Vullikanti, VBI and CS, K. Channakesava (Ericsson)

Objective /Goals

Impacts Highlights

Sponsors: NSF, DTRA

Efficient techniques for detection, analysis and control of the spread of malware

New approaches to capture heterogeneous and mobile networks Realistic multi-scale models of mobile

networks – proximity based (BlueTooth) and Internet

Scalable tools for simulation and analysis of malware spread on very large networks

Spread of malware in Activity-based networks

Spread of malware in Random mobility models

New ways to control the spread, detect the onset and finding vulnerable devices

Mobility matters - Realistic networks are structurally different than simple random networks.

Structural differences affect how malware spreads.

Underscores the importance of studying human mobility and its impact on wireless networks

EpiCure: HPC-modeling framework that can represent and study malware over urban scale (with millions of nodes), time varying and heterogeneous networks

1000 times faster than prior approaches Approach based on epidemics on social

contact networks, BlueTooth specific states abstracted out Approximation error less than 5%

Published in: IPDPS’11, Simutools’09, BioWire’09

Time

Sp

ace

Internet worms

Human worms

DaysSeconds

Mobileworms

Threat of malware increasing with smart phones and mobile devices

Challenges: multiple-scales, self forming networks, need to model user behavior

NLOS Cooperative Sensor LocalizationReza Monir Vaghefi and R. Michael Buehrer

Objective /Goals

Highlights

Cooperative time-of-arrival-based sensor localization in a non-line-of-sight (NLOS) environment is investigated.

Cooperative sensor localization plays an important role in indoor networks where GPS is limited.

However, indoor networks suffer from NLOS propagation which degrades localization accuracy significantly.

We assume that the estimator is able to discriminate NLOS connections from line-of-sight (LOS) connections.

We introduce a novel semidefinite programming (SDP) approach for cooperative localization which exploits NLOS connections to enhance the accuracy of localization.

0 1 2 3 4 5 60

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Norm of location error [m]

CD

F

ML-LOSSDP-MSDP-LOSIPPMML-LOS+NLOSLLS-LOS+NLOS

NLOS connections add a large positive bias to the measurements which usually causes large estimation error.

If the estimator is optimal (ML estimator), neglecting NLOS connections is the best approach.

If the estimator is sub-optimal (e.g., SDP and LLS), using NLOS connections improves the localization accuracy.

A novel SDP (SDP-M) technique was introduced in which both LOS and NLOS connections are used to improve the accuracy of location estimation

Ontology-based Spectrum Access Policies for Policy-based Cognitive Radios

Objective

Impact and ApplicationsHighlights

PI: Jung-Min “Jerry” Park

• Design of a policy reasoner that processes ontology-based spectrum policies.

• Achieving the full benefits of ontology-based spectrum policies: interoperability, flexible knowledge representation, self-awareness

• Overcoming challenges of ontology-based policy reasoning.

o Meeting the real-time processing requirements of a cognitive radio.

o Handling continually varying state information of a CR.

• Our ontology-based policy reasoner:o Uses SWRL (Semantic Web Rule Language)

rules to represent spectrum policies. o Capable of handling all ontology operations,

including ontology-consistency checking and ontology information editing.

o Uses a tableau-based algorithm to carry out ontology reasoning.

o Uses a modified Rete algorithm to process spectrum policies and evaluate transmission requests.

• Design and implementation of a policy reasoner that processes ontology-based spectrum access policies.

• Provide more expressive spectrum policies for policy-based cognitive radios such as XG radio.

• Using our implementation, we have carried out a number of experiments. Based on the experiments results, we provide insights and guidelines for designing ontology-based spectrum access policies.

Architecture of the ontology-based policy reasoner.

CAREER: Opportunistic Routing in Multihop and Multirate Wireless Networks

Objectives

• Basic idea is to study how effective OR can improve the network capacity and how the protocols should be designed to take full advantage of the broadcast nature of the wireless medium and the spatial diversity of network topology.

• A theoretical framework on the computation of performance bounds of OR.

• Design and implementation of distributed and efficient OR algorithms and protocols.

Highlights

• A theoretical framework on capacity of opportunistic routing in multi-rate, multi-radio, multi-channel, and multi-hop wireless networks

• Energy-efficient geographic opportunistic routing in lossywireless ad hoc and sensor networks

• Opportunistic broadcast protocols with network coding in vehicular networks

• Reliable broadcast with network coding in wireless mesh networks

• Secure and fast coordination schemes for opportunistic routing

Opportunistic Routing: tries to make every transmission useful. Actual path is determined after the transmission. A best node among the nodes having received the packet is used to forward the packet. Sponsored by National Science Foundation, Sep 2008~ Aug 2013

Impacts/Applications

• OR is a general idea to combat unreliable wireless links.

• Can be applied to various wireless technologies, such as multi-rate, multi-radio, multi-channel, in various networking scenarios, such as multi-hop wireless networks, mobile wireless networks, for both unicast and broadcast communications.

• Can also be combined with other techniques such as network coding to further improve the performance.

P=0.7

P=0.4

P=0.5

P=0.5

P=0.6

P=0.7

PI: Wenjing Lou

Outdoor Cognitive Radio Network Testbed PIs: Tamal Bose, Haris I. Volos, and Jeff Reed

Objectives/Goals

ImpactsHighlights

• Enhance our indoor CORNET testbed with outdoor capabilities• Test dynamic scenarios with indoor and outdoor connectivity for applications such as wireless distributed computing• Perform spectrum activity measurements/studies• Investigate the coexistence of cognitive networks with radar systems

Proposed nodes map

Period of performance: July 2012 –December 2013

Sponsor: Office of Naval Research

• 15 nodes placed throughout Virginia Tech’s campus• 2 mobile nodes (in addition to the 10 mobiles nodes that Virginia Tech recently acquired)• The nodes will be based on the Universal Radio Peripheral (USRP) by Ettus research

• Each node will be equipped with a GPS disciplined oscillator for precise time stamping of received and/or transmitted signals

• Virginia Tech will offer a one of a kind in the world cognitive radio testbed consisting of 48+ indoor nodes, 12 mobile nodes, and 15 outdoor nodes• It will support current and future research programs from Department of Defense funded and the National Science Foundation• Like CORNET, the new testbed will be made available to Virginia Tech’s nationwide and worldwide collaborators

Power Fingerprinting Monitor

Pow

er T

race

Software Modules

t

Power Fingerprinting in SDR and CR Integrity AssessmentJeffrey. H Reed and Carlos R. Aguayo Gonzalez

Objective /Goals

Applications and ImpactHighlights

Period of performance: Sep 2009 – Oct 2011Sponsor: National Science Foundation

Match real-time power

traces against trusted

signatures

Malicious intrusions cause deviations from the signatures

• Fine-grained, anomaly detection on the processor’s power consumption

• Independent, physically separated sensor

• Embedded or external monitor

• References from trusted implementations or models provide a baseline for comparison

• Obtained before deployment during characterization phase in controlled environment

• Feature extraction and classification

• Quantitative measure of trust: How closely the observed power consumption matches a reference

• Demonstrate the feasibility of PFP on commercial platforms

• Identify the best discriminatory features to uniquely identify software execution for the target platform

• Develop optimal classifiers that include multiple features

• Investigate methodologies to profile software modules and extract reference signatures

• Evaluate expected intrusion detection performance

• Develop a framework to perform regulatory policy enforcement using power fingerprinting

• Effective mechanism to enforce regulatory policy in SDR and CR

• Detect unauthorized modifications in certified radios

• Intrusion detection and integrity assessment in wireless and embedded systems

• Cyber security for smart phones, industrial control

• Independent of platform, OS, and application

• Effective against zero-day attacks

• Inherent protection for monitor due to physical separation

• Minimal overhead on the target device

SDR Technology DevelopmentJeffrey H. Reed, Carl Dietrich, Peter Athanas

Objective /Goals

Impacts Highlights

Period of Performance 2010-2012Sponsor: US Department of Defense

To investigate and develop SDR related technology

• Multiband RF front end

• APIs for SDR on heterogeneous processing platforms

• Implementations of selected processing blocks

• Addition of flexible control interface to selected processing blocks

Multiband RF front end, related hardware, and APIs will facilitate SDR research

SDR implementations/components for use in research

Control interface will allow run-time configuration and distributed SDR applications

SDR Hardware and APIs under development

APIs demonstrated in signal processing application

SDR processing blocks identified for use with flexible control interface

Security & Robustness of TD-LTE for Public-Safety CommunicationsInvestigators: C. Shahriar, S. Sodagari & T. Charles Clancy

Objective /Goals

The prime candidate for 5G is DSA-enabled TD-LTE where OFDM is used as ‘the’ waveform. But OFDM based waveform is not used in defense up to now for security issues. Securing DSA and OFDM will promote new standard in public-safety communications. This will open up door for further research.

Highlights

The ultimate goal of this research project is to develop a secure and robust communications system that can be used in sensitive and mission critical situation such as defense and public-safety network.

Period: August 2011 to December 2012 Sponsor: Security and Software Engineering

Research Center (S2ERC)

Survey physical layer security issues including waveform and DSA securities.

Develop power efficient jamming such as pilot jamming and pilot nulling.

Develop method to mitigate the pilot jamming by randomizing pilot locations.

Develop new DSA strategy named Hop on Stop instead of existing Do No Harm.

Impacts and Applications

NEXT GENERATION SECURE, SCALABLE COMMUNICATION NETWORK FOR SMART GRID

PI: R. Michael Buehrer, Co-Pis: Jeffrey H. Reed and Haris I. Volos

Objectives/Goals

ImpactsHighlights

• The core objective is to conduct research, development, and demonstrations leading to next generation secure and scalable communication network for smart grid

Virginia Tech will investigate:• Spreading code and multiple access design for

CDMA mesh networks;• Interference/jamming mitigation techniques;• Multiuser detection techniques; and• Combined communication and position location

using spread spectrum waveforms.

Period of performance:February 2011 – December 2012

Sponsor:Oak Ridge National Laboratory (ORNL)

• A new Smart Grid for the generation, transmission, and distribution of electricity is under development• The Smart Grid is depended on wireless technologies for facilitating its operations• The wireless technologies used must be secure and robust•ORNL has a patented Hybrid Spread Spectrum Direct Sequence and Frequency Hopping waveform• Virginia Tech will assist ORNL with applying their waveform to the needs of the Smart Grid

• Meeting the goals of this project, would provide better support to enhance the cyber-security of communication and control systems specific-to smart grid to efficiently perform the production and distribution of electricity.

Spectral Coexistence of Radar and Communication SystemsShabnam Sodagari

Objective /Goals

Applications and Impacts Highlights

Sponsor: ONR

H1

H2

H3

1) Enabling both radar and comm to interleave transmission in time/frequency holes

2) Enabling both to coexist on the same central frequency (e.g. 2-4 GHz for S-band) in geographical proximity

Null Spaced Based Radar Transmission Radar acquires and updates REM

Cooperative: Cognitive Pilot Channel (CPC) Non-cooperative: semi-blind techniques, e.g., HMM

Radar maps its signals onto null space of interference channel

Fusion of Wireless Communication and Radar Sensing

Providing more bandwidth as required by the national broadband plan

Spectrum sharing with commercial communication systems e.g., LTE, WiMAXwith mitigated interference to both systems and closer proximity

SPIRITDr. Jules White, Paul Miranda, Thaddeus Czauski

Objective/Goals

• Using location as a method of authentication

• Improved indoor localization for mobile devices

• Using a device’s geographic location as an access control mechanism

Applications

•Safeguarding data based on location

•Reconfiguring devices on a per-location basis

•Providing users relevant information and reminders based on their location

Highlights

• Implementation and testing underway

Impacts

• Cyber-physical systems

• Location aware computing

• Software agents

Healthcare Commercial Government

Study on Tactic Effectiveness Mitigating DSA Radio ThreatsMarc Lerch, Chris Jennette

Objective /Goals

Impacts By implementing a variety of unique and independent tactics in unison, a DSA radio may be able to mitigate the threat of all but the most powerful and versatile Primary User Emulation (PUE) jammers attacks.

ApplicationsA web-based DSA model that takes in parameters such as number of radios, radio tactics, number of interferers (jammers), number of channels, type of jamming, 3 dimensional geography, and pathloss calculations will accurately simulate the performance of DSA radios in a threat-infested environment. By use of a plug-in to Virginia Tech’s CORNET these simulation results can then be validated against actual radio performance.

These results can then be used to contribute to the shaping of Secure Spectrum Access Policy Research.

Highlights

Dynamic Spectrum Access (DSA) technologies offer a unique opportunity to increase spectral efficiency in heterogeneous environments by dynamically selecting a viable over-the-air configuration within policy, spectral, and environmental constraints. Early research has suggested that these systems could be attacked by adversaries impersonating license holders, causing secondary users to vacate the band based on their regulatory requirements. This program studies an approach that identifies a potential attack through an understanding of the modus operandi of a primary user emulation and mitigate the potential attack by implementing a variety of countermeasures, which exploit the attacker’s limits while not disrupting license holders.

Four types of threats are studied:•Sweeping Jammer - randomly selects channels independent of the radio frequency environment. •Reactive Jammer - selects the channel with the strongest signal power during each channel sensing period. •Barrage Jammer - selects one channel without considering the radio frequency environment and stays there. •Herding Jammer - randomly selects channels independent of the radio frequency environment while never occupying one “destination” channel. Proposed tactics:Access new bands, adaptive thresholding, change occupancy period, change channel selection strategy, force periodic channel changes.

Period: August 2011 to July 2012 Sponsor: DARPA with Shared Spectrum as Prime

Throughput Optimization of Cooperative Relaying in Wireless Networks

Investigators: Thomas Hou, Scott F. Midkiff, Hanif D. Sherali

Objectives

• Exploit cooperative communications (CC) in a multi-hop ad hoc network

• Optimize network level throughput via cooperative relaying

• Understand potential and performance limits of cooperative relay networks

Impacts• Advancing a key technology to increase

throughput and reliability of wireless networks

• Offer solutions at the network layer for cooperative relay networks

• Develop new mathematical models and tools

• Develop cross-disciplinary course materials for wireless networking

Highlights

• Optimal use of network coding (NC) in cooperative relay networks

• Relay node selection for throughput maximization

• Explore performance bounds of cooperation in multi-hop relay networks

Graphic or Photo

Period of performance: 7/2011 – 6/2014Sponsor: NSF

Trust for Reconfigurable Embedded and Cyber-physical SystemsLee W. Lerner and Cameron D. Patterson, {lwl, cdp}@vt.edu

Objective / Goals

Highlights

9/1/9 – 8/31/12

1. Generalization of domain-specific (Cognitive Radio) security project

2. Supervisory and application software is in the critical path of control

• Difficult to verify and susceptible to malware

• Has influence over configurable hardware datapaths/controllers

3. Little effort has been made to secure configurable controllers which manage cyber-physical interactions

• When compromised, provide ability to corrupt system or hijack physical processes resulting in damage or unavailability

• Assembled from 3rd-party modules with varying degrees of trust

1. Protects process controller configurations from compromised or erroneous supervisory and privileged software

2. Protects physical processes from embedded controller software and IP modules that are untrusted or poorly understood

• Provides decision logic for systems implementing redundancy

1. Instilled in the system through automatic synthesis of security policies and expected operations represented as assertions

• Built from and integrated with existing design languages and tools to allow seamless adoption with system designers

• Process control reliability projection derived from attributes of system models developed during a model-based design flow

• The ROTH is updatable, but only with physical access

1. Configurable hardware-assisted rule enforcement (CHARE)

• A proactive, domain-independent design-for-security, -trust, and -reliability (DFSTAR) methodology for embedded systems

• Trusted specification guards provide control flow assurances through enforcement of system requirements

2. Process control violation predictors

• Embedded process models enable projection of future controller commands and process state

• Control sequences resulting in process instability are rejected

3. Root-of-trust hardware (ROTH) for cyber-physical systems

• CHARE specification guards and violation predictors provide a ROTH for embedded controllers in cyber-physical systems

• Enforcement of system expectations is a tractable defense against cyber attacks on critical infrastructure

Root-of-trust Hardware (ROTH)

1. CHARE protects the most safety-critical part of the cyber system

2. Specification guards are digital interlocks ensuring reliable control

Using Coordinated Multi-point TX/RX (CoMP) to Implement Self Healing in LTE-A

Objectives Impact and Applications

PI: Jung-Min “Jerry” Park

• Maximizing the throughput inside the outage area

• Achieving proportional fairness among users in the outage area

• Preventing overloading neighboring cells through achieving even distribution of outage area load

• Minimizing human intervention in case of cell outage which results in minimizing OpEX

• Minimizing the required time to “heal” the network

Introduction

• Self-Organizing Networks (SONs) is one of LTE-A promising features to reduce OpEX

• When outage happens, it is the responsibility of surrounding cells to “heal” the outage area

• CoMP is one of the promising features in LTE-A to enhance the performance for edge users

• CoMP is categorized into Joint Processing (JP), and Coordinated Scheduling/Beamforming (CS/CB)

Wireless Assessment of the Gigapark Sites inthe Mid-Atlantic Broadband Cooperative Footprint

PIs: Haris I. Volos and Jeff Reed

Objectives/Goals

ImpactsHighlights

• To assess the quality of service of each of the major wireless carriers for each of the 64 GigaParks sites in south/central Virginia. The service metrics are:

• Signal strength• Voice quality• Data throughput

•Only 2G/3G services were considered, i.e., no LTE or WiMax (they were not available in the area at the time)The MBC parks

Period of performance: Jul. – Dec. 2011Sponsor: Mid-Antlantic Broadband Cooperative

(MBC)

• Evaluated the voice and data services of six carriers: AT&T, nTelos, Sprint, T-Mobile U.S. Cellular and Verizon

• At each site we performed • A drive test while placing automated

voice calls• Data throughput tests

• U.S. Cellular had the best data speeds and signal levels, however, served fewer sites

• Verizon serviced most of the sites• T-Mobile had the poorest performance in all

metrics

• The data collected will help MBC in its mission to assist in economic revitalization of the areas survey by attracting businesses to the area•We developed a custom Android application, called VT-CELLAR (CELLular Automatic Recoded) which automatically recorded the signal strength and placed test calls

Wireless Distributed and Social Computing : Architecture, Protocols & PrimitivesC. Kuhlman, M. Khan, M. Marathe, A. Vullikanti,

Virginia Bio-informatics Institute and Dept. Computer Science

Objective /Goals

Impact Highlights

Sponsors: DTRA, NSF, DARPA

Model driven architectures that scale and

exploit context (what, when, where and who)

Bio-inspired paradigms

Efficient and robust distributed primitives Time, message and energy efficient

Ability to deal with dynamic networks

HPC-based modeling environments to design

and analyze WDC systems Networks with 100,000 nodes and more

• DASim: An HPC-based modeling environment to study distributed protocols, primitives &contagion in wireless social networks

• Algorithms and Statistical techniques to measure performance of large wireless networks

• CoSMO: Cognitively-inspired architecture for distributed sensing, situation assessment and control

• Efficient distributed algorithms, e.g. Topology controlMinimum spanning tree (MST) Broadcast

Challenges:• Interference• Dynamic topology• Energy is crucial• Time critical • Must be scalable to

large number of nodes

New methods for opportunistic offloading that reduce backbone traffic by 50%Methods for targeting nodes to improve the scope and efficiency information diffusionModels suggest ways to develop wireless architectures that can support energy efficient network primitives

Results published in: IEEE JSAC’09, TPDS’09, INFOCOM’08, WINET’11 MONET.

Future wireless networks cannot be effectively designed, analyzed and controlled in isolation without taking into account the social context

Wireless Distributed Computing (WDC)PI: Jeff Reed

Objectives/Goals

ImpactsHighlights

Develop•Preliminary mathematical models•System level architecture;•Standardized APIs•Algorithms and protocols

Demonstration•Using USRPs (Virginia Tech)•Using WNaN radios (BBN)

Period of performance:August 2010 – August 2011

Sponsor:DARPA

• WDC enables low cost/capability devices to perform demanding tasks to

• Improve computational latency• Improve battery life

• Improves security, a captured device doesn’t have all the data and code• Redundancy, no single point of failure• The wireless channel and battery levels pose a challenge•Resource management and optimization is required

• Replace expensive nodes with many inexpensive and possibly disposable nodes• Use currently available, but under utilized SDR resources to reduce processing time• New applications can be developed based on the WDC architecture and APIs• The code developed will be available as open source enabling further development of the methods implemented

Wireless Grid Innovation Testbed (WiGiT)Tamal Bose, Haris Volos

Objective /Goals

Impacts Highlights

• Develop a wireless grid innovation testbedcoordinated by Virginia Tech and Syracuse University with academic, government and corporate partners.• Spur growth of new companies and transformative wireless grid applications.• Facilitate people-to-people, people-to-resources, and people-to-facilities interactions.• Create a Virtual organization by using edgeware and cognitive radio.• Encompass wide-range and geographically dispersed activities and group through VO partnership.

Create four 4 new applications

Emergence notification system

Wireless distributed computing

Distributed Learning

4 Universities joined as VO

2 new graduate courses, 1 in proposal

Several media reports

Analyzing and Improving OFDM for Acoustic Channels

Sponsored by

Currently open

Principal Investigator A. A. (Louis) Beex

Student Researcher

Avik Dayal

Project Duration Ongoing

OFDM (orthogonal frequency division multiplex) communication is useful for an underwater acoustic channel as it can combat the severe multipath that occurs in littoral environments. We aim to also mitigate the wideband Doppler effect that occurs between moving platforms.

Our wideband OFDM communication system uses the over-the-air channel. Due to room acoustics, the channel is subject to multipath and – when microphones are moving, with respect to the transmitting speaker – subject to Doppler warping. Our efforts are directed towards making improvements in the overall design of the system, such as different approaches to packet and symbol synchronization and to modeling and estimation of the Doppler variations. Experiments show that in the presence of substantial time-varying Doppler effects the performance of synchronization algorithms deteriorates rapidly.

Recently we have achieved coarse synchronization of packets that have undergone severe Doppler and multipath distortion, as seen below, and illustrated by the frequency shifts undergone by sinusoids in the pre- and post-ambles.

Increasing Utilization of Legacy Channels/Spectrum

Sponsored by

Currently open

Principal Investigator A. A. (Louis) Beex

Student Researcher

TBD

Project Duration Ongoing

When a particular channel is assigned to a particular user, that channel can – presumably – not be used by someone else. The particular utilization of such a channel can then be quite low, as it remains idle until it is needed. Considering the dearth of bandwidth, or number of channels, it would be good to find a way to use available bandwidth that is not currently in use by the legacy, or primary, user(s). Of course, a secondary user must not interfere with the primary user. Also, the existence of the secondary user should be transparent to the legacy user, i.e. the latter need not be concerned with the existence of the former. It is the secondary user who needs to insure that the channel is only used when it is not in primary use. When the legacy user uses a known modulation format, this creates some a priori information that can be used to facilitate a secondary user. For example, a communications burst often consists of a known synchronization pattern followed by the modulated burst carrying the actual information. The secondary user must simultaneously sense the synchronization pattern in order to vacate the channel before interfering with the primary user. The detection of the synchronization pattern must therefore be quick. If necessary, the detection process for the start of the primary burst can have a large probability of false alarm to insure that the channel is being vacated in time. Once the channel is being used by the primary user, the sensing task of the secondary user becomes one of detecting when the modulation portion of the legacy burst is over. Generally a different detector is needed to detect the modulation portion of the signal, especially when the signal bandwidth approaches the channel bandwidth. To detect the end of the primary burst, the probability of false alarm (deciding modulation has ended when in fact it has not) must be very low while now the probability of detection need not be very high. We are looking into various modulation detection approaches. Another aspect of this project is to determine to what extent collisions of the legacy and secondary communications bursts impact the performance of the legacy system.

Modeling & Recognition of Transients in Speech

Sponsored by

Currently open

Principal Investigator A. A. (Louis) Beex

Student Researcher

Areg Baghdasaryan

Project Duration On hold

If we can determine appropriate models for the transients in speech vowels, then the onset of the transient can be detected and the specific vowel identified or recognized. Characterization of the transients in speech is expected to be beneficial for the enhancement of speech coder quality. Another application of the identification task is to be able to associate the particular vowel with the particular person speaking that vowel. This aspect of the project is aimed at speaker identification/verification/authentication.

Mitigation of Time-Varying Narrowband Interference and Multipath from CPM

Sponsored by

Currently open

Principal Investigator A. A. (Louis) Beex

Student Researcher

TBD

Project Duration Ongoing

Mitigation of time-varying narrowband interference is the first goal of this project. Initial efforts were based on predictive modeling with time-varying coefficients. As CPM signals are finitely correlated, this required decorrelating time-varying prediction. The time-varying prediction system causes signal distortion, expressed in ISI, which can be compensated for because the predictive system parameters are continually estimated and thus “known.” We have shown that this approach can effectively mitigate the interference over some range of SIR (signal-to-interference ratio), to the point where the raw SER (symbol-error-rate) is sufficiently good for a Viterbi decoder to work (TVLPEF 20 dB curve in Fig. 1). Towards extending the usable SIR range, we recently concentrated on a time-varying model for the narrowband interference (i.e. no longer predictive). This approach was indeed successful in expanding the range of performance (over the -20 to -5 dB SIR range in Fig. 1), even though this came at an increased expense in terms of computational effort.

Fig. 1. SER vs. SIR for ES interference mitigation from a CPM signal.

Future efforts are towards joint modeling of the interference and the signal, as well as finding the most practical approaches towards (approximately) solving the resulting optimization problems. The latter speaks to the eventual goal of designing and implementing an effective algorithm in hardware.

Nonlinear Effects in Adaptive Filtering

Sponsored by

Currently open

Principal Investigator A. A. (Louis) Beex

Researcher/Collaborator

Takeshi Ikuma (LSU)

Student Researcher Tamoghna Roy

Project Duration

Ongoing

Adaptive algorithms for noise cancellation, prediction, and equalization may exhibit nonlinear effects when operating in environments with strong narrowband interference present. The nonlinear effects are expressed by time-varying or dynamic behavior of the weights of the transversal adaptive filter, even in a wide-sense stationary environment, and performance that exceeds that of the Wiener filter with the corresponding structure. We’ve shown – and analyzed – that the mean of the adaptive equalizer filter weights does not correspond to the Wiener weights. The mean-square error of the adaptive equalizer is significantly smaller than that of the fixed Wiener equalizer, as illustrated below. Further performance improvement was obtained by our recent BLMS algorithm.

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-12

-11

-10

-9

-8

-7

-6

-5

-4

-3

normalized stepsize, Lmax

stea

dy-s

tate

MS

E,

J (

dB)

J(v

)

J(vw

)

BLMS (tr)

LMS (tr)BLMS (dd)

LMS (dd)

Jw

For our BLMS equalizer, BER performance may under some scenarios be better by as much as several orders of magnitude relative to the FIR Wiener equalizer. We are analyzing the new algorithm in order to explain how the improved performance is achieved, and investigating further applications for NB interference dominated scenarios.

Robust Automatic Modulation Classification

Sponsored by

Currently open

Principal Investigator A. A. (Louis) Beex

Student Researcher

TBD

Project Duration Ongoing

Knowing whether spectrum is being used and to some extent in what fashion – without actually demodulating a received signal – depends on being able to classify what type of modulation is seen, if any. The main approaches are based on time-spectral features and on statistical features. The latter category often exploits any cyclostationarity properties the signal may exhibit. We are looking at the latter approaches and – in particular – towards extending their range of performance, i.e. lowering the SNR required for achieving a certain high probability of correct classification at or below some low level of missed detection. Use of the robust estimator has led to detection, and classification, at lower SNR, for the same or better probability of detection and false alarm as the non-robust approach, as illustrated in the left figure below.

The robust approach is also less sensitive to a mismatch between the receiver bandwidth and the signal bandwidth. Importantly, our results indicate that the robust approach achieves the same performance as the non-robust approach using only one tenth the number of observations, as illustrated in the figure above, on the right. CPM signals exhibit cyclo-stationarity of different order, depending on the modulation indices involved. A more practical detector for CPM signals has been designed that is based on second-order cyclo-stationarity only. Next we aim to extend the SNR range of performance for the detection of CPM by making the latter estimator robust. Further extension will then aim to widen the class of modulations to include CPM.