Research

Special thanks to our sponsors:


My research involves stochastic design and control of sensing, communication, information processing, and learning in modern engineered systems. This covers broad theoretical questions in interactive learning and information acquisition as well as a variety of practical implementations from computer vision to service drones to wireless networks.  On the theoretical front, I am most concerned with the problem of information acquisition and interactive learning where the cost of data collection and/or labeling can be substantially reduced. Here the challenge is to deal with imperfect and noisy data as well as inconsistent response from labelers. Here our objective has been to 1) develop algorithms that  acquire the most informative features with the  minimum query complexity and 2)  design queries that account for the uncertainty and inconsistency of humans in the loop.  On the more practical front, I am interested to apply our developed algorithms in the following three application domains: 1) next generation wireless networks , 2) service drones, and even 3) computer vision. 

In my research group, we often prove theorems but we also build and test our theoretical findings when possible (graduate applicants, please refer to this link!). Here, I try to break down our work in (overlapping) areas:

Information Acquisition-Utilization and Controlled Sensing:
Dynamic Tracking with Imperfect Observation
Active Sequential Hypothesis Testing and Communications with Feedback
Bayesian Noisy Active Learning

Stochastic Control and Optimization of Networks:
Integrated Data Center Networking
Cognitive Networking
Optimal Routing in Wireless Mesh and Multi-hop Networks

Communications with Application Constraints:
Delay Sensitive Communications
Information Theoretic Secrecy in Presence of Feedback
Integrated Communication and Control

For a detailed discussion of our work in each of these categories, see below (sorted temporally):

Information Acquisition and Utilization:

This is where stochastic control theory meets signal processing and information theory. Here we consider a generalization of active sequential hypothesis testing (see below) and Hidden Markov Models (HMM). Much similar to an HMM problem, we assume a Markovian stochastic process whose state is available to us only through partial (noisy) observations. In classical HMM problems, given some knowledge of the conditional distribution of observations given the true state, we are often interested in filtering or smoothing our observations to arrive at estimate of the true state of the process.  However, in an information acquisition problem,  very much like our treatment of active hypothesis testing, we assume that the statistics of these observations can be controlled and tuned actively and dynamically by our decision maker and beyond filtering the state estimate is used to utilize certain resources. The applications are wide and include the very fundamental problem of 

To top

Dynamic Tracking with Imperfect Observation
There is a variety of applications where the decision maker needs to track a partially observable stochastic process. The examples include:

Enhanced Spectrum Access:

Javidi, T.; Krishnamachari, B.; Qing Zhao; Mingyan Liu; , "Optimality of Myopic Sensing in Multi-Channel Opportunistic Access. IEEE International Conference on Communications, 2008. ICC '08, 19-23 May 2008 [here]

Ahmad, S.; Mingyan Liu; Javidi, T.; Qing Zhao; Krishnamachari, B.; , "Optimality of Myopic Sensing in Multichannel Opportunistic Access," Information Theory, IEEE Transactions on , vol.55, no.9, pp.4040-4050, Sept. 2009 [here] 

P. Mansourifard, T. Javidi, and B. Krishnamachari. Optimality of Myopic Policy for a Class of Monotone Affine

Restless Multi-Armed Bandits. in Proceedings of IEEE Conference on Decision and Control (CDC), December, 2012.

Search and Tracking with Measurement-dependent Noise:

S. Chiu and T. Javidi. Sequential Measurement-Dependent Noisy Search. Proceedings of IEEE Information Theory Workshop (ITW), September 2016

A. Lalitha and T. Javidi. Reliability of Sequential Hypothesis Testing Can Be Achieved by an Almost-Fixed-Length Test. in Proceedings of the IEEE International Symposium on Information Theory (ISIT), July 2016.

M. Naghshvar and T. Javidi. Two-Dimensional Visual Search. in Proceedings of International Symposium Information Theory (ISIT), July 2013

M. Naghshvar and T. Javidi. Rate–Reliability Tradeoff in Two-Dimensional Visual Search. in Proceedings of Iran Workshop on Communication and Information Theory (IWCIT), May 2013 [Invited]

And even joint source-channel coding:

T. Javidi and A. Goldsmith. Dynamic Joint Source–Channel Coding with Feedback. in Proceedings of International Symposium Information Theory (ISIT), July 2013

As one expects this area of research is very much motivated by our work on active hypothesis testing. 

To top

Active Sequential Hypothesis Testing and Communications with Feedback:

This is where stochastic control theory meets information theory: Consider the classic “I-SPY: Can You See What I See” game in which a child is given a page and is asked to identify the location of an object against a crowded background. Given our eyes’ inherent non-uniform information gathering across the field of vision, i.e. reduced vision perception and precision away from the center of gaze onto the periphery, the following question arises: How do our brains best control the uncertain visual information collection process and refine our belief in this and similar environments? The problem of dynamical control of an information state hardly remains as a singular childhood preoccupation with the I-SPY game nor is limited to the vision application domain. Related problems arise in the context of sequential design of scientific experiments, sensor activation in health monitoring, and even communication over noisy channels with feedback. My work with my former graduate student Mohammad Naghshvar considers a broad spectrum of applications in cognition, communications, design of experiments, and sensor management. In all of these applications, a decision maker is responsible to control the system dynamically so as to enhance his information in a speedy manner about an underlying phenomena of interest while accounting for the cost of communication, sensing, or data collection. Furthermore, due to the sequential nature of the problem, the decision maker relies on his current information state to constantly (re-)evaluate the trade-off between the precision and the cost of various actions: 

M. Naghshvar, T. Javidi, and K. Chaudhuri. Noisy Bayesian active learning. in Proceedings of 50th Annual Allerton Conference on Commununication, Control and Computation, October 2012

M. Naghshvar and T. Javidi. Extrinsic Jensen--Shannon divergence with application in active hypothesis testing. in Proceedings of IEEE International Symposium on Information Theory (ISIT), July 2012

M. Naghshvar and T. Javidi. Active Sequential Hypothesis Testing: Sequentiality and Adapativity Gains. in Proccedings of Conference on Information Sciences and Systems (CISS). Princeton, NJ. 2012 [here]

M. Naghshvar and T. Javidi. Performance Bounds for Active Sequential Hypothesis Testing. to appear in Proceedings of IEEE International Symposium on Information Theory (ISIT), August 2011 [here]

M. Naghshvar and T. Javidi. Information utility in active sequential hypothesis testing. In Proceedings of Forty-eighth Annual Allerton Conference on Communication, Control, and Computing, Monticello, Illinois, 2010 [here]

M. Naghshvar and T. Javidi. Active M-ary Sequential Hypothesis Testing. in Proceedings of IEEE International Symposium on Information Theory (ISIT), 2010 [here]

For more detailed treatment, you can check out our more detailed manuscripts on arxiv:

M. Naghshvar and T. Javidi, Active Sequential Hypothesis Testing. Annals of Statistics. Vol. 41, No. 6, December 2013

(see the version on arxiv: [here] )

M. Naghshvar and T. Javidi, Sequentiality and Adaptivity Gains in Active Hypothesis Testing. To appear in IEEE Journal of Selected Topics in Signal Processing, Vol. 7,  No. 5, October  2013. 

(see the version on arxiv: [here] )

Interestingly, we have shown that the problem of channel coding over a memoryless noisy channel (point to point as well as multiple access) with feedback is nothing but a special case of the active sequential hypothesis problem analyzed in certain asymptotic regimes:

Mohammad Naghshvar, Tara Javidi, and Michele Wigger. Extrinsic Jensen–Shannon Divergence:

Applications to Variable-Length Coding. [Arxiv]

M. Naghshvar and T. Javidi. Optimal reliability over a DMC with feedback via deterministic sequential coding. in Proceedings of IEEE International Symposium of Information Theory and Applications (ISITA), October 2012 [here]

M. Naghshvar, M. Wigger, and T. Javidi, Optimal reliability over a class of binary-input channels with feedback. in Proceedings of IEEE Information Theory Workshop (ITW), September 2012 [here]

E. Ardestanizadeh, M. A. Wigger, Y.H. Kim, and T. Javidi. Linear-Feedback Sum-Capacity for Gaussian Multiple Access Channels. IEEE Transactions on Information Theory, Issue 1, volume 58. January 2012 [here]

E. Ardestanizadeh, M. A. Wigger, Y.H. Kim, and T. Javidi. Linear sum capacity for Gaussian multiple access channel with feedback, Proceedings of IEEE International Symposium on Information Theory (ISIT), 2010

M. Naghshvar and T. Javidi. Variable-length coding with noiseless feedback and finite messages. in Proceedings of Asilomar Conference on Signals, Systems, and Computers, November 2010 [here]

Ardetsanizadeh, E.; Javidi, T.; Young-Han Kim; Wigger, M.A.; , "On the sum capacity of the Gaussian multiple access channel with feedback,"  47th Annual Allerton Conference on Communication, Control, and Computing, 2009 [here]

To top

Bayesian and Non-Bayesian Noisy Active Learning
An important and closely related problem is the problem of noisy active learning. In this context, we have applied our work on extrinsic Jensen--Shannon divergence to this problem to obtain strong guarantees. In particular, we have addressed the following problems:

S. Yan, K. Chaudhuri and T. Javidi. Active Learning from Noisy and Abstention Feedback, Allerton Conference on Communication, Control and Computing, 2015 

M. Naghshvar, T. Javidi, and K. Chaudhuri. Bayesian Active Learning With Non-Persistent Noise. IEEE Transactions on Information Theory. Vol 61. Issue: 7. July 2015 

M. Naghshvar, T. Javidi, and K. Chaudhuri. Noisy Bayesian active learning. in Proceedings of 50th Annual Allerton Conference on Commununication, Control and Computation, October 2012


Stochastic Control and Optimization of Networks:

To top

Integrated Data Center Networking:

My earlier work on switch and network scheduling:

T. Javidi, R Magill, and T. Hrabik. A High Throughput Scheduling Algorithm for a Buffered Crossbar Switch Fabric. In Proceedings of IEEE International Conference on Communications, vol. 5, pp. 1586-1591, 2001

M. Naghshvar, H. Zhuang, and T. Javidi. A General Class of Throughput Optimal Routing Policies in Multi-hop Wireless Networks. IEEE Transactions on Information Theory, Volume 58 , Issue 4, pp 2175 - 2193, 2012

became the basis of my interactions with the Center of Integrated Access Networks (CIAN) whose research thrusts includes the integration of the optical switch technology in intra-data center networking. While optical communication technology can provide extremely fast and efficient link for data transmission, the adoption of optical technology in large scale networking is impeded by the general inefficiency of writing and buffering in optical domain. To address the issue, we conceptualized the overall non-blocking non-buffering end-end network as a generalized switch (with less than full bisection bandwidth). 

We have showed the power of this network-wide generalization of switching by showing that the problem of end-end scheduling in data-center networking can be adapted to the switching cost and loss of duty-cycle. We developed a scheduling algorithm that extended all previous work on switch scheduling to the case of flow management for  optical switching:

C. Wang, T. Javidi, G. Porter. End-to-End Scheduling for All-Optical Data Centers. in Proceedings of IEEE International Conference on Computer Communication (INFOCOM), April 2015

Subsequently, along with a postdoc of mine, Tugcan Aktas, we showed that the proposed architecture can easily be supported as a hybrid wireless/optical networking solution:

T. Javidi, C. Wang, T. Aktash. A Novel Data Center Network Architecture with Zero in-Network Queuing. Proceedings of the 2015 IEEE International Symposium on Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks (WiOpt), May 2015 [invited]

T. Aktas, C. Wang, T. Javidi. WiCOD: Wireless Control Plane Serving an all-Optical Data Center. Proceedings of the 2015 IEEE International Symposium on Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks (WiOpt), May 2015

To top

Optimal Routing in Wireless Mesh Networks:

I have also used closely related techniques in stochastic optimization and approximation to address the problem of routing in ad hoc and mesh wireless networks. As a part of my thesis work, I investigated the sensitivity of opportunistic routing to errors in the underlying Bayesian model. Recently, with my graduate students at UCSD, we have extended this work to multi-packet scenarios in which reinforcement learning techniques are used to achieve the optimal performance in the absence of information about the underlying (stochastic) model of the network. Furthermore, to address the issue of congestion diversity we have combined ideas from opportunistic routing and back-pressure to arrive at throughput optimal policies with good delay performance.

T. Javidi and D. Teneketzis. Sensitivity Analysis for Optimal Routing in Wireless Ad Hoc Networks in Presence of Error in Channel Quality Estimation. IEEE Transactions on Automatic Control, August 2004 [here]

A. Bhorkar, M. Naghshvar, T. Javidi and B. Rao, An Adaptive Opportunistic Routing Scheme for Wireless Ad-hoc 

Networks. in IEEE Transactions on Networking, August 2011 [here]

M. Naghshvar, H. Zhuang, and T. Javidi. A General Class of Throughput Optimal Routing Policies in Multi-hop Wireless Networks, accepted for publication in IEEE Transactions on Information Theory, 2011 [here]

P. Gupta and T. Javidi, Towards Throughput and Delay Optimal Routing for Wireless Ad-Hoc Networks. In 

Proceedings of Asilomar Conference on Signals, Systems, and Computers, November 2007 [here]

M. Naghshvar and T. Javidi, Opportunistic Routing with Congestion Diversity in Wireless Multi-hop Networks.

INFOCOM, 2010 Proceedings IEEE , vol., no., pp.1-5, 14-19 March 2010 [here]

A. Bhorkar. Javidi. No Regret Routing for ad-hoc wireless networks. Conference Record of the Forty Fourth Asilomar Conference on Signals, Systems and Computers (ASILOMAR), vol., no., pp.676-680, 7-10 Nov. 2010 [here]

Recently, I have gotten interested in the practical implementation of these solutions in 802.11 mesh networks. In particular, we have set up a testbed in Calit2 building consisting of 12 or so wifi nodes equipped with Atheros cards to test the ideas above on, and evaluate the practicality of our theoretical ideas.

A. Bhorkar, T. Javidi, and A. Snoeren. Achieving congestion diversity in wireless ad-hoc networks. Proceedings IEEE INFOCOM, vol., no., pp.521-525, 10-15 April 2011 [here]

To top

Cognitive Networking:

As the demand for mobile/wireless services as well as the complexity and the diversity of networks and devices continue to grow, there is a need for developing affordable and scalable means for effectively utilizing the available resources to deliver the complex applications and services of the future. It this context, cognitive networking has come to the forefront of wireless networking research. While, cognitive radio technology was first envisioned around the notion of spectrum agility, with the ever increasing popularity of the “smart” devices -not only equipped with multiple wireless interface cards but also with significant storage and computational capabilities- the once dream-like notions of sensing and network computations, adaptability, and learning are now generalized across the network and protocol stack. In particular, we have looked at two aspects of cognitive networking. The first set of work attempts to addresses the exploitation versus exploration tradeoff in context of opportunistic sensing/access, while the second class of our publications deal with the problem of network/topology information in wireless multihop routing. 

Javidi, T.; Krishnamachari, B.; Qing Zhao; Mingyan Liu; , "Optimality of Myopic Sensing in Multi-Channel Opportunistic Access. IEEE International Conference on Communications, 2008. ICC '08, 19-23 May 2008 [here]

Ahmad, S.; Mingyan Liu; Javidi, T.; Qing Zhao; Krishnamachari, B.; , "Optimality of Myopic Sensing in Multichannel Opportunistic Access," Information Theory, IEEE Transactions on , vol.55, no.9, pp.4040-4050, Sept. 2009 [here] 

A. Bhorkar, M. Naghshvar, T. Javidi and B. Rao, An Adaptive Opportunistic Routing Scheme for Wireless Ad-hoc 

Networks. in IEEE Transactions on Networking, August 2011 [here]

A. Bhorkar. Javidi. No Regret Routing for ad-hoc wireless networks. Conference Record of the Forty Fourth Asilomar Conference on Signals, Systems and Computers (ASILOMAR), vol., no., pp.676-680, 7-10 Nov. 2010 [here]

P. Mansourifard, T. Javidi, and B. Krishnamachari. Optimality of Myopic Policy for a Class of Monotone Affine Restless Multi-Armed Bandits. in Proceedings of IEEE Conference on Decision and Control (CDC), December, 2012

A third class of problems arise in the context of distributed learning and estimation. Here with my collaborator, Anand Sarwate, we have combined techniques from consensus literature with stochastic approximation and analysis to arrive at learning distributions across the network:

A.D. Sarwate, T. Javidi, Distributed Learning of Distributions via Social Sampling, Accepted for publication in IEEE Transactions on Automatic Control. [Arxiv]

To top

Delay Sensitive Communications and Information Transmission:

The problem of data transmission over noisy discrete memoryless channels is considered when the arrival of data is bursty and is subject to a delay deadline. In this setting, the errors experienced by the source of information, concatenated with an infinite buffer and a noisy channel, are caused by either erroneous decoding at the receiver or violation of the delivery deadline. It is intuitive, then, that there is a trade off between reliability over the channel and timely delivery of information. Briefly revisiting the difficulty in quantifying the above trade-off in its classical Shannon Theoretic context, we have addressed this problem in particular instances with and without feedback:

R. Swamy and T. Javidi. Delay analysis of block coding over a noisy channel with limited feedback. Conference 

Record of the Forty Second Asilomar Conference on Signals, Systems and Computers (ASILOMAR) Oct. 2008 [here]

R. Swamy and T. Javidi. Delay analysis of block coding over a noisy channel with limited feedback. Proceedings of 

IEEE International Symposium on Information Theory (ISIT). July 2009 [here]

Unfortunately, the above analysis quickly become analytically intractable and not so elegant beyond the simple intuitions they provide. We have also taken advantage of the high SNR analysis of outage-limited channels to go around this difficulty. Hence, the focus of the work becomes to characterize the error performance of the overall system in the high SNR regime:

S. Kittipiyakul, P. Elia, and T. Javidi. High-SNR analysis of outage-limited communications with bursty and delay-limited information IEEE Trans. Inf. Theory, v. 55, no. 2, pp. 746-763, Feb 2009.  [ .pdf]

S. Kittipiyakul and T. Javidi, Optimal operating point for MIMO multiple access channel with bursty traffic IEEE Trans. on Wireless Communications, v. 6, no. 12, Dec. 2007. [ .pdf]

S. Kittipiyakul and T. Javidi, Optimal operating point in MIMO channel for delay-sensitive and bursty traffic ISIT'06, July 2006. [ .pdf]

This characterization motives novel queuing theoretic work on a many source large deviation analysis of queues with batch service and/or multi-queue systems with Max-Weight scheduling policies. In particular, we use a recent extension of contraction principle, due to Garcia, to arrive at the appropriate large-deviation of a queue concatenated by a Max-weight scheduled multi-user channel, which is of independent interest. 

Subramanian, V.G.; Javidi, T.; Kittipiyakul, S. Many-Sources Large Deviations for Max-Weight Scheduling. Information Theory, IEEE Transactions on , vol.57, no.4, pp.2151-2168, April 2 [here]

To top

Stochastic Optimization in Multi-server Queueing Systems

Many problems of information processing in networks are best modeled as a generalization of the classical multi-server multi-queue problem with scheduling constraints [1-5]. These problems are often mapped to a family of Markov decisions problems (MDPs) and partially observed MDPs (POMDPs).   By identifying the optimal server allocation policies analytically or their salient structures, we can bypass the computational complexity and significantly improve the performance in terms of delay, blocking probabilities, or revenue/cost. My contributions in this area build on tractable models and are analytical in nature. 

T. Javidi, N. Song, and D. Teneketzis. Expected Makespan Minimization in Two Interconnected Queues on Identical Parallel Machines. Journal of Probability in Engineering and Information Science, vol. 15, pp 409-443, 2001[pdf] 

An extended version can be found here.

T. Javidi and D. Teneketzis. An Approach to Connection Admission Control in Single-hop Multi-service Wireless Networks with QoS Requirements.  IEEE Transactions on Vehicular Technology, July 2003 [pdf]

N. Ehsan and T. Javidi, Delay Optimal Transmission Policy in a Wireless Multi-access Channel. IEEE Transactions on Information Theory, Volume 54,  Issue 8, pp. 3745 – 3751, August 2008 [pdf]

S. Kittipiyakul and T. Javidi, Delay-Optimal Server Allocation in Multi-Queue Multi-Server Systems with Time-Varying Connectivities. IEEE Transactions on Information Theory. Volume 55, Number 5, April 2009 [pdf]

T. Javidi. Cooperative and Non-cooperative Resource Sharing in Networks: A Delay Perspective. IEEE Transactions on Automatic Control, Volume 53, Number 9, pp 2134-2142, October 2008 [pdf]

Subramanian, V.G.; Javidi, T.; Kittipiyakul, S. Many-Sources Large Deviations for Max-Weight Scheduling. Information Theory, IEEE Transactions on , vol.57, no.4, pp.2151-2168, April 2 [here]

But more importantly, they enable simple-to-implement heuristics, e.g., practical subcarrier allocations for OFDMA, or new buffered architecture for throughput optimal scheduling for switching:

S. Kittipiyakul and T. Javidi, Resource allocation in OFDMA with time-varying channel and bursty arrivals, IEEE Communications Letters, v. 11, no. 9, pp. 708-710, Sept 2007 [pdf]

T. Javidi, R Magill, and T. Hrabik. A High Throughput Scheduling Algorithm for a Buffered Crossbar Switch Fabric. In Proceedings of IEEE International Conference on Communications, vol. 5, pp. 1586-1591, 2001 [.pdf]

To top

Communications under Application and/or Secrecy Constraints

With the success of wireless communication techniques many novel applications and issues have emerged. One such application is communication under secrecy constraints. The maximum rate of information transfer from a transmitter, Alice, to a receiver, Bob, secret from an eavesdropper, Eve, known as secrecy capacity between Alice and Bob, has been a topic of study for multiple decades. We have studied the secrecy capacity of a communication systems where there is an additional rate-limited secret feedback channel from Bob to Alice. The work sheds light on the significance of secret feedback channel and its role in increasing secrecy. 

Ardestanizadeh, E.; Franceschetti, M.; Javidi, T.; Young-Han Kim. Wiretap Channel With Secure Rate-Limited Feedback. IEEE Transactions on Information Theory, vol.55, no.12, pp.5353-5361, Dec. 2009 [here]

Another area of interest is in the domain of coordinated multi-vehicle motion planning and control. Prior to joining UCSD, I collaborated with Prof. K. Morgansen at the University of Washington, Seattle, to develop an integrated communication and control architecture for coordinated control of multi-vehicles. Along with our graduate students, D. Klein and P. Lee, we established a trade-off between communication costs and cost of control in the control of vehicles with Kuromoto under-actuated controllers.  We further extended the heading control problem by considering a randomized scheme for achieving consensus in a network. 

D. Klein, P.Lee, K. Morgansen and T. Javidi, "Integration of Communication and Control using Discrete Time Kuramoto Models for Multivehicle Coordination over Broadcast Network," IEEE Journal on Selected Areas in Communications, v. 26, no. 4, pp. 695-705, May 2008. [pdf]

R. Pagliari, M.E. Yildiz, S. Kirti, K. Morgansen, T. Javidi, and A. Scaglione. A Simple and Scalable Algorithm for Alignment in Broadcast Networks. in IEEE Journal on Selected Areas in Communications, August 2010 [here]

To top

Applications of Microeconomic Theory to Network Design

Wired and wireless networks can be viewed as economies where the goods are bandwidth, buffer size, and power. The goals and characteristics of network design and management also match the requirements and characteristics usually associated with economic markets in many aspects such as scalability, efficiency, and decentralization. My former PhD student, Jennifer Price, and I used simple and basic ideas in microeconomics to design and analyze wireless and wired networking protocols. 

J. Price and T. Javidi. Decentralized Rate Assignments in a Multi-Sector CDMA Network. IEEE Transactions on 

Wireless Communications, December 2006 [pdf]

J. Price and T. Javidi, Leveraging Downlink for Efficient Uplink Allocation in a Single-Hop Wireless Network. IEEE Transactions on Information Theory, v. 53. no. 11, Nov 2007. [pdf]

J. Price and T. Javidi, Distributed Rate Assignments for Simultaneous Interference and Congestion Control in CDMA-Based Wireless Networks. IEEE Transactions on Vehicular Technology, Volume 57, Issue 3, May 2008 [pdf]

J. Price and T. Javidi. Network Coding Games with Unicast Flows. IEEE Journal on Selected Areas in Communications, Volume 26,  Issue 7, pp. 1302 – 1316, September 2008 [pdf]