Recent Research: Using AI for Wireless Networks
I am currently leading a DARPA-funded three-year project on intelligent spectrally efficient wireless communications. The project is part of the second DARPA Spectrum Collaboration Challenge (SC2).
Our BAM! Wireless team spans six research groups at Purdue and Texas A&M. The goal is to augment different layers of the communication stack with
intelligent decision making for accomplishing a set of tasks that ranges from identifying interference sources and their modulation schemes to
better recognition of channel conditions. Further, we aim to investigate how intelligence can leverage the value of coordination in
wireless networks.
Our approach is focused on exploring appropriate deep neural network architectures for the considered machine learning tasks, as well
as for real-time control (deep reinforcement learning). We also plan to consider “shallow” learning techniques like kernel-based methods
and graph-based semi-supervised learning for scenarios where the availability of labeled samples is limited.
Recent Research: Cloud-Based Wireless Networks
The goal of this project is to advance the use of the cloud in next generation wireless networks through ideas that are backed by
information-theoretic analysis. The availability of the cloud can enable new approaches that were previously seen infeasible.
Decisions of cell associations as well as transmission schedule and coding schemes can be taken in a centralized fashion with global
knowledge of the network in hand. Further, learning of the network topology and channel state information can be done through
a centralized cloud coordinator.
This project consists of three parts: The first part is dedicated to a study of infrastructural networks with the assumption of
knowledge of the channel state information at all nodes. The second part is dedicated to the study of heterogeneous and Ad-hoc networks with
no channel state information available at the transmitters. The third part is dedicated to the study of centralized strategies for
learning the network topology and channel state information.
Postdoctoral Research: Big Data Analytics
I am working on a study of semi-supervised learning in collaboration with the research group of Prof. Antonio Ortega.
The study consists of two parts, analyzing the optimal structures to represent the sampled data and analyzing optimal reconstruction methods to recover the labels of unlabeled data from few labeled examples.
Understanding fundamental limits in this setting is important to derive benchmarks for the performance of practical algorithms in the context of big data analytics where large sets of unlabeled samples are available with few labeled examples that are provided through expert feedback.
In the accomplished part of this project, we analyzed the recently introduced method of
Band-limited Interpolation of Graph-signals (BIG) for semi-supervised learning. By analyzing the convergence of the
bandwidth of the graph signal, we showed that as the number of samples increases, the decision boundary recovered through
the BIG approach is closely related to the low density separation problem. By analyzing the distribution of the graph Laplacian
eigenvalues, we derived the required number of labels by the BIG approach, and compared it with spectral clustering methods
to highlight the value of labelled samples when clustering is not a good solution.
M.Sc. and Ph.D. Research: Interference Management and Security of Wireless Networks
My M.Sc. and Ph.D. studies have been focused on analyzing the information theoretic limits of wireless communication scenarios that have practical significance. The goal of my M.Sc. work was to find novel ways to exploit features of the wireless channel in order to achieve physical-layer secure communication regardless of the computational capabilities of the adversary. We developed a randomized transmission scheme to achieve information theoretic secrecy using two-way communication, and inspired by the fact that a binary erasure channel has larger capacity than that of an erroneous counterpart with the same probability of erased/corrupt symbols.
The goal of my Ph.D. thesis is to provide an information-theoretic characterization of the potential gains offered by Coordinated Multi-Point (CoMP) transmission in interference networks. We consider the problem in which each user's message can be available at more than one transmitter in the network. The selection of a message assignment reflects the setup of backhaul links in a cellular downlink scenario. Given limited backhaul capacity, we analyze Degrees of Freedom (DoF) optimal assignment of messages and transmission scheme in interference networks with various architectures.
Theses Advised
Deep Neural Network Architectures for Modulation Classification. Xiaoyu Liu. March 2018.
Book
Interference Management in Wireless Networks: Fundamental Bounds and the Role of Cooperation, V. V. Veeravalli, A. El Gamal, Cambridge University Press, Feb. 2018.
Patent
LIDAR: Lifelong-learning-based Intelligent, Diverse, Agile and Robust architecture for network attacks detection. A. El Gamal, A. Elghariani, A. Ghafoor, Provisional filed, Feb. 2020.
Journal Publications
Gradient-based Adversarial Deep Modulation Classification with Data-driven Subsampling, J. Yi, A. El Gamal, submitted, Mar. 2021.
Mitigating Gradient-based Adversarial Attacks via Denoising and Compression, R. Mahfuz, R. Sahay, A. El Gamal, submitted, Mar. 2021.
A Number Theoretic Approach for Fast Discovery of Single-Hop Wireless Networks, T. Seyfi, A. Mohamed, A. El Gamal, submitted, Dec. 2020.
Deep Learning for Direction of Arrival Estimation in MIMO Radar Systems via Emulation of Large Antenna Arrays, A. Ahmed, U. Thanthrige, A. El Gamal, A. Sezgin, Accepted at the IEEE Communications Letters, Jan. 2021.
Deep Learning for Frame Error Prediction using a DARPA Spectrum Collaboration Challenge (SC2) Dataset, S. Jameel, A. Mohamed, X. Zhang, A. El Gamal, submitted, Dec. 2020.
Efficient Training of Deep Classifiers for Wireless Source Identification using Test SNR Estimates. [Modulation Classification Code]. [Channel Identification Code]. X. Wang, S. Ju, X. Zhang, S. Ramjee, A. El Gamal, accepted for publication at the IEEE Wireless Communications Letters, Apr. 2020.
Non-Intrusive Detection of Adversarial Deep Learning Attacks via Observer Networks. [Code]. K. Sivamani, R. Sahay, A. El Gamal, IEEE Letters of the Computer Society, vol. 3, no. 1, pp. 25-28, Jan-Jun. 2020.
Ensemble Wrapper Subsampling for Deep Modulation Classification. [Code]. S. Ramjee, S. Ju, D. Yang, X. Liu, A. El Gamal, Y. C. Eldar. submitted, May 2020.
Joint Uplink-Downlink Cooperative Interference Management with Flexible Cell Associations, M. Singhal, T. Seyfi, A. El Gamal, accepted for publication at the IEEE Transactions on Communications, May 2020.
Fundamental Limits of Dynamic Interference Management with Flexible Message Assignments and Separate Deep Fading Block Coding, T. Seyfi, Y. Karacora, A. El Gamal, IEEE Transactions on Information Theory, vol. 66, no. 2, pp. 1193-1212, Feb. 2020.
Ensemble Noise Simulation to Handle Uncertainty about Gradient-based Adversarial Attacks. [Code]. R. Mahfuz, R. Sahay, A. El Gamal. submitted, Feb. 2020.
Efficient Wrapper Feature Selection using Autoencoder and Model Based Elimination. [Code]. S. Ramjee, A. El Gamal. submitted, May 2020.
Vision Paper: Grand Challenges of Resilience: Autonomous System Resilience through Design and Runtime Measures, S. Bagchi, V. Aggarwal, S. Chaterji, F. Douglis, A. El Gamal, J. Han, B. J. Henz, H. Hoffman, S. Jana, M. Kulkarni, F. X. Lin, K. Marais, P. Mittal, S. Mou, X. Qiu, G. Scutari. accepted for publication at the IEEE Open Journal of the Computer Society, Jun. 2020.
A Computationally Efficient Method for Defending Adversarial Deep Learning Attacks. [Code]. R. Sahay, R. Mahfuz, A. El Gamal, submitted, Dec. 2019.
On the Design and Analysis of Optimal Wireless Caching Schemes with Placement Cost and Unlimited Memory, Y. Alhassoun, F. Alotaibi, A. El Gamal, H. El Gamal, accepted for publication at the IEEE Wireless Communications Letters, Apr. 2020.
Fast Deep Learning for Automatic Modulation Classification. [Code]. S. Ramjee, S. Ju, D. Yang, X. Liu, A. El Gamal, Y. C. Eldar, IEEE Machine Learning for Communications Emerging Technologies Initiatives, Jan. 2019.
Degrees of Freedom in Wirless Interference Networks with Cooperative Transmission and Backhaul Load Constraints, M. Bande, A. El Gamal, V. V. Veeravalli, IEEE Transactions on Information Theory, vol. 65, no. 9, pp. 5816-5832, Sep. 2019.
A Single Coin Monetary Mechanism for Distributed Cooperative Interference Managament, A. El Gamal, H. El Gamal, IEEE Wireless Communications Letters, vol. 8, no. 3, pp. 757-760, Jun. 2019.
A Sampling Theory Perspective of Graph-Based Semi-Supervised Learning, A. Anis, A. El Gamal, A. S. Avestimehr, A. Ortega, IEEE Transactions on Information Theory, vol. 65, no. 4, pp. 2322-2342, Apr. 2019.
Fundamental Limits of Non-Coherent Interference Alignment via Matroid Theory, N. Naderializadeh, A. El Gamal, A. S. Avestimehr, IEEE Transactions on Information Theory, vol. 63, no. 10, pp. 6573-6586. Oct. 2017.
Interference Channels with CoMP: Degrees of Freedom, Message Assignment, and Fractional Reuse, A. El Gamal, V. S. Annapureddy, V. V. Veeravalli, IEEE Transactions on Information Theory, vol. 60, no. 6, pp. 3483-3498, Jun. 2014.
The Two-Way Wiretap Channel: Achievable Regions and Experimental Results, A. El Gamal, O. O. Koyluoglu, M. A. Youssef, H. El Gamal, IEEE Transactions on Information Theory, vol. 59, no. 12, pp. 8099-8114, Dec. 2013.
Degrees of Freedom of Interference Channels with CoMP Transmission and Reception, V. S. Annapureddy, A. El Gamal, V. V. Veeravalli, IEEE Transactions on Information Theory, vol. 58, no. 9, pp. 5740-5760, Sep. 2012.
Conference Publications
Knowledge Distillation for Wireless Edge Learning. A. Mohamed, S. Jameel, A. El Gamal, Submitted, Mar. 2021.
A Provably Convergent Information Bottleneck Solution via ADMM. T. Huang, A. El Gamal, Submitted, Feb. 2021.
Data-driven Analysis of Turbulent Flame Images. R. Roncancio, J. Kim, A. El Gamal, J. P. Gore, AIAA Science and Technology Conference 2021, Jan. 2021.
Efficient Coded Caching with Limited Memory, Y. Alhassoun, F. Alotaibi, A. El Gamal, H. El Gamal, Allerton Conference on Communications, Control, and Computing, Sep. 2019.
Deep Learning for Interference Identification: Band, SNR, and Sample Selection. [Code]. X. Zhang, T. Seyfi, S. Ju, S. Ramjee, A. El Gamal, Y. C. Eldar, IEEE International Workshop on Signal Processing advances in Wireless Communications (SPAWC), Jul. 2019.
Towards Jointly Optimal Placement and Delivery: To Code or Not to Code in Wireless Caching Networks, Y. Alhassoun, F. Alotaibi, A. El Gamal, H. El Gamal, IEEE International Symposium on Information Theory (ISIT), Jul. 2019.
Wyner's Network on Caches: Combining Receiver Caching with a Flexible Backhaul, E. Lampiris, A. El Gamal, P. Elia, Accepted at IEEE International Symposium on Information Theory (ISIT), Mar. 2019.
Combatting Adversarial Attacks through Denoising and Dimensionality Reduction: A Cascaded Autoencoder Approach, R. Sahay, R. Mahfuz, A. El Gamal, Conference on Information Sciences and Systems - Johns Hopkins University, Mar. 2019.
Optimal Cell Associations and Degrees of Freedom of Locally Connected Interference Networks with Message Passing Decoding, M. Singhal, A. El Gamal, IEEE International Symposium on Information Theory, Vail, Colorado, Jun. 2018.
Deep Neural Network Architectures for Modulation Classification. [Code]. X. Liu, D. Yang, A. El Gamal, Asilomar Conference on Signals, Systems, and Computers, Nov. 2017.
The Role of Transmitter Cooperation in Linear Interference Networks with Block Erasures, Y. Karacora, T. Seyfi, A. El Gamal, Asilomar Conference on Signals, Systems, and Computers, Nov. 2017.
Joint Uplink-Downlink Cell Associations for Interference Networks with Local Connectivity, M. Singhal, A. El Gamal, Allerton Conference on Communication, Control, and Computing, Oct. 2017.
Topological Interference Management: Linear Cooperation is not useful for Wyner's Networks, A. El Gamal, IEEE International Symposium on Information Theory (ISIT), Aachen, Germany, Jun. 2017.
Cloud-Based Topological Interference Management: A Case with No Cooperative Transmission Gain, A. El Gamal, IEEE International workshop on Signal Processing advances in Wireless Communications (SPAWC), Edinburgh, Jul. 2016.
Cell Associations that Maximize the Average Uplink-Downlink Degrees of Freedom, A. El Gamal, IEEE International Symposium on Information Theory (ISIT), Barcelona, Jul. 2016.
When Does an Ensemble of Matrices with Randomly Scaled Rows Lose Rank?, A. El Gamal, N. Naderializadeh, A. S. Avestimehr, IEEE International Symposium on Information Theory (ISIT), Hong Kong, Jun. 2015.
Flexible Backhaul Design with Cooperative Transmission in Cellular Interference Networks, M. Bande, A. El Gamal, V. V. Veeravalli, IEEE International Symposium on Information Theory (ISIT), Hong Kong, Jun. 2015.
Asymptotic Justification of Band-Limited Interpolation of Graph Signals for Semi-Supervised Learning, A. Anis, A. El Gamal, A. S. Avestimehr, A. Ortega, International Conference on Acoustics, Speech and Signal Processing (ICASSP), London, Apr. 2015.
Topological Interference Management with just Retransmission: What are the Best Topologies, N. Naderializadeh, A. El Gamal, A. S. Avestimehr, accepted for publication in proceedings of the International Conference on Communications (ICC), Jan. 2015.
Flexible Backhaul Design and Degrees of Freedom for Linear Interference Channels, A. El Gamal, V. V. Veeravalli, IEEE International Symposium on Information Theory (ISIT), Hawaii, Jul. 2014.
Dynamic Interference Management, A. El Gamal, V. V. Veeravalli, Asilomar Conference on Signals, Systems, and Computers, Nov. 2013.
Degrees of Freedom of Locally Connected Interference Channels with Cooperating Multiple-Antenna Transmitters, A. El Gamal, V. S. Annapureddy, V. V. Veeravalli, IEEE International Symposium on Information Theory (ISIT), MIT, Jul. 2012.
Degrees of Freedom of Locally Connected Interference Channels with Coordinated Multi-Point (CoMP) Transmission, A. El Gamal, V. S. Annapureddy, V. V. Veeravalli, International Conference on Communications (ICC), Ottawa, Jun. 2012.
On Optimal Message Assignments for Interference Channels with CoMP Transmission, A. El Gamal, V. S. Annapureddy, V. V. Veeravalli, 46th Annual Conference on Information Sciences and Systems (CISS), Princeton, Mar. 2012.
Degrees of Freedom of Cooperative Interference Networks, V. S. Annapureddy, A. El Gamal, V. V. Veeravalli, IEEE International Symposium on Information Theory (ISIT), Saint Petersburg, Aug. 2011.
Degrees of Freedom of the K-user Interference Channel with Transmitter Cooperation, V. S. Annapureddy, A. El Gamal, V. V. Veeravalli, IEEE International Symposium on Information Theory (ISIT), Austin, Jun. 2010.
New Achievable Secrecy Rate Regions for the Two Way Wiretap Channel, A. El Gamal, O. O. Koyluoglu, M. A. Youssef, H. El Gamal, IEEE Information Theory Workshop (ITW), Cairo, Jan. 2010.
Randomization for Security in Half-Duplex Two Way Gaussian Channels, A. El Gamal, M. A. Youssef, H. El Gamal, IEEE Global Communications Conference (Globecom), Hawaii, Dec. 2009.
|