Master 2

Machine learning, communications, and security


Overview


This second-year program is devoted to information sciences and is centered around the question “How to process, transmit, store, and protect information efficiently?”

The cross-disciplinary master MICAS starts with foundational courses in information processing, and builds around three main areas:
● Machine Learning
● Communications Theory and Technologies
● Security

The courses are organized into twelve teaching units (UEs) and involve professors from three laboratories of Institut Polytechnique de Paris (IP Paris), namely, from Telecom Paris, Telecom SudParis, and ENSTA. These courses are provided in English over a single semester from September to February. An internship follows thereafter for a minimum period of four months.

MICAS offers mathematically minded students a strong background on both fundamentals of information processing and related technologies. This prepares them to later pursue an academic career or a career in the industry in the fields of data science and/or communication technologies.

Language of instruction: English
ECTS: 60
Oriented: Research
Duration: 1 year
Courses Location: Telecom Paris and Telecom SudParis, IP Paris Campus at Palaiseau


Educational objectives


This cross-disciplinary program aims to provide students a mathematical and practical background on information processing concepts and tools with state-of-the-art applications related to learning, communications and security. Upon completion of the program, students will be able to identify complex problems in these fields, understand their interactions, and exhibit solutions with performance and limitation assessments.

The program will qualify students to pursue careers in a wide array of domains, whether in industry or in academia.


Program structure


The program is structured into two semesters of 60 ECTS.
The first semester (September-January) is dedicated to foundational courses in information processing and more advanced courses in the fields of Machine Learning, Communications and Security. These courses are organized into 12 teaching units corresponding to 36 ECTS, including lectures and seminars (CMs), exercises (TDs) and labs (TPs). They are taught in English and assessed through a combination of assignments, quizzes, and end-of-semester projects and exams.
The second semester is devoted to an internship of minimum four months (24 ECTS) in an academic laboratory or R&D industry. The internship is evaluated based upon a master thesis report and an oral defense.

The teaching units (UEs) are categorized into 4 groups of courses. For the first group on “Fundamentals in Information processing”, 1 elective course can be chosen between F3.1 on information theory and F3.2 on communication theory. The teaching units are listed below and their detailed presentation is given thereafter.

Fundamentals (F) in Information Processing (10.5 ECTS = 105 hours)
● F1: Introduction to Convex Optimization
● F2: Introduction to Statistical Inference
● F3.1: Introduction to Information Theory
● F3.2: Introduction to Communication Theory

Machine Learning (ML) (9 ECTS = 90 hours)
● ML1: Statistical Learning
● ML2: Sequential Decision-Making
● ML3: Deep Learning

Communications (COM) (10.5 ECTS = 105 hours)
● COM1: Multi-user Communications
● COM2: Advanced Wireless Communications
● COM3: Advanced Coding

Security (SEC) (6 ECTS = 60 hours)
● SEC1: Cryptography
● SEC2: Secure Communications

UE F1: Introduction to Convex Optimization

Coordinators:
Hadi Ghauch, Associate Professor, Telecom Paris
Michele Wigger, Professor, Telecom Paris

Volume: 30 hours 3 ects
Hours: Lectures: 22.5h, Exercises: 6h
Assessment: 2 Assignments and 1 Final Exam
Language: English

Objectives:
The course provides an in-depth study of the mathematical theory of convex and non-convex optimization, with an emphasis on learning applications.

Outcomes:
On completion of the course students should be able to:
● Understand the theory and algorithms for convex and non-convex optimization
● Apply a wide array of optimization tools

Prerequisite:
● Basic notions of linear algebra and calculus

Syllabus
● Convex functions, convex sets, strong convexity
o Convex problems, equivalence between problems
● Lagrange duality, optimality conditions for convex problems, KKT conditions, Slater’s condition, strong duality
● Algorithms to solve convex optimization problems
o gradient methods, Newton’s method, Frank-Wolfe algorithm
o accelerated algorithms: Nesterov Acceleration, Heavy-ball method
o proximal methods: proximal gradient
o Convergence analysis of algorithms
o stochastic optimization
● Decomposition methods: primal and dual decomposition methods, ADMM
● Distributed optimization:
o non-convex optimization
o non-convex functions/problems,
o non-convex optimization methods: successive approximation methods, coordinate descent methods,
o and block coordinate descent methods
o relaxations for non-convex problems
o discrete/combinatorial optimization problems
relaxation for combinatorial optimization: Lagrange relaxation, Dantzig Wolfe relaxation, semi-definite relaxation
● Applications: convex/non-convex problems in machine learning and wireless communications

Bibliography:
● S. Boyd and L. Vandenberghe, “Convex optimization”, 2004.
● D. Bertsekas, “Nonlinear programming”, 3rd Ed, 1999

UE F2: Introduction to Probability and Statistics

Coordinators:
Philippe Ciblat, Professor, Telecom Paris
Aslan Tchamkerten, Associate Professor, Telecom Paris

Volume: 45 hours 4.5 ects
Hours: Lectures: 30h, Exercises: 12h
Assessment: Final Exam
Language: English

Objectives:
The course presents basic statistical concepts needed for communications and learning.

Outcomes:
On completion of the course students should be able to:
● Understand certain fundamental limitations to inference
● Apply the tools to communication and learning problems

Prerequisite:
● Real analysis

Syllabus
● Probability
o probability space
o independence
o random variables
o LLN
o central limit theorem
o tails
o concentration
o subgaussianity

● Detection and estimation
o Hypothesis testing, performance bounds
o Detection: max a posteriori detector
o Estimation: bayesian approach (mean a posteririo) performance bounds (Cramer-Rao Bound)
o Estimation: deterministic approach (maximum likelihood, least square), asymptotic analysis.
o unbiasedness and equivariance
o Data reduction

Bibliography:
● S. M. Kay, “Fundamentals of Statistical Signal Processing: Estimation Theory”, Vol. 1, Prentice Hall, 1993.
● P. Moulin and V. V. Veeravalli, “Statistical Inference for Engineers and Data Scientists”, Cambridge, 2019.
● H. Vincent Poor, “An Introduction to Signal Detection and Estimation”, 2nd Edition, Springer, 1998.
● H. L. Van Trees, “Detection, Estimation and Modulation Theory”, Wiley, 2001.

UE F3.1: Introduction to Information Theory

Coordinators:
Olivier Rioul, Professor, Telecom Paris
Aslan Tchamkerten, Associate Professor, Telecom Paris

Volume: 30 hours 3 ects
Hours: Lectures: 18h, Exercises: 9h
Assessment: Final Exam
Language: English

Objectives:
The course addresses the two fundamental information questions, namely “How to compress data efficiently?” and “How to transfer data efficiently?”

Outcomes:
On completion of the course students should be able to:
● Understand the basics of Shannon’s information theory
● Evaluate optimal trade-offs and Shannon limits for a particular source or channel model

Prerequisite:
● Real analysis
● Introduction to Probability and Statistics

Syllabus
● Source and Channel Models
● Entropy, Relative Entropy (Divergence), Conditional Entropy, Mutual Information
● Data Processing Inequality, Fano’s Inequality
● Asymptotic Equipartition Property, Typical Sequences
● Source Coding, Universality
● Channel Coding, Capacity
● Rate-Distortion
● Joint Source-Channel Coding
● Gambling

Bibliography:
● T.M. Cover and J. A. Thomas, “Elements of information theory”, John Wiley & Sons, 2012.
● A. El Gamal and Y.-H. Kim, “Network Information Theory”, Cambridge university press, 2011.

UE F3.2: Introduction to Communication Theory

Coordinators:
Philippe Ciblat, Professor, Telecom Paris
Mireille Sarkiss, Associate Professor, Telecom SudParis

Volume: 30 hours 3 ects
Hours: Lectures: 19.5h, Exercises: 6h, Labs: 3h
Assessment: 2 Assignments and 1 Final Exam
Language: English

Objectives:
The course provides the main mathematical tools and technologies for communication systems.

Outcomes:
On completion of the course students should be able to:
● Design a communication system given some system requirements
● Analyze the performance and compare systems
● Understand the principles of error correcting codes

Prerequisite:
● Introduction to Probability and Statistics
● Linear algebra

Syllabus
● Baseband/carrier signals
● Multipath Channel modeling
● Detection Theory: Maximum A Posteriori (MAP) and Maximum Likelihood (ML) detector
● Gaussian channel receiver
● Frequency-Selective channel
o Linear and nonlinear receivers: Zero Forcing (ZF), Minimum Mean Square Error (MMSE), Decision Feedback Equalizer (DFE)
o Orthogonal Frequency Division Multiplexing (OFDM)
● Application to 4G
● Introduction to channel coding
● Linear block coding
● Galois Fields
● Cyclic coding

Bibliography:
● D. Tse, “Fundamentals of Wireless Communications”, 2005.
● Goldsmith, “Wireless Communications”, 2005.
● S. Lin and D. J. Costello, “Error Control Coding: Fundamentals and Applications”, 2004.

UE ML1: Statistical Learning

Coordinators:
Hadi Ghauch, Associate Professor, Telecom Paris
Aslan Tchamkerten, Associate Professor, Telecom Paris

Volume: 30 hours 3 ects
Hours: Lectures: 24h, Exercises: 4.5h
Assessment: Final Exam
Language: English

Objectives:
The course provides a mathematical introduction to supervised and unsupervised statistical learning.

Outcomes:
On completion of the course students should be able to:
● Identify the different types of inference problems (supervised/unsupervised)
● Apply algorithms
● Know certain intrinsic limitations to inference problems

Prerequisite:
● Linear algebra
● Introduction to Convex Optimization
● Introduction to Probability and Statistics

Syllabus
● Statistics vs. learning approaches
● Supervised learning
o Binary classification
o Empirical risk
o VC theory
o Boosting
o Support Vector Machine (SVM)
o Prediction with expert advice
o Linear methods for regression
● Unsupervised learning
o Cluster analysis
o K-Means
o Principal components

Bibliography:
● Felipe Cucker and Steve Smale, « On the Mathematical Foundations of Learning », Bulletin of the American Mathematical Society, 2002.
● Trevor Hastie, Robert Tibshirani, Jerome Friedman, « The Elements of Statistical Learning », 2008.
● Philippe Rigollet, “Mathematics of Machine learning », MIT lecture notes, 2015.
● Martin Wainwright, High-Dimensional Statistics A Non-Asymptotic Viewpoint, Cambridge University Press, 2019.
● P. Bishop, “Pattern recognition and machine learning”, 2006.
● L. Bottou, F. Curtis and J. Norcedal, “Optimization Methods for Large-Scale Machine Learning”, SIAM Rev., 60(2), 223–311, 2018.

UE ML2: Sequential Decision Making

Coordinators:
Philippe Ciblat, Professor, Telecom Paris
Mireille Sarkiss, Associate Professor, Telecom SudParis

Volume: 30 hours 3 ects
Hours: Lectures: 18h, Exercises: 4.5h, Labs: 6h
Assessment: 2 Assignments and 1 Final Exam
Language: English

Objectives:
The course provides Markovian Decision Process Theory and shows how it can be applied in many settings. An introduction to Deep Reinforcement Learning (DRL) is also conducted.

Outcomes:
On completion of the course students should be able to:
● Identify a Markov Decision Process setting
● Check assumptions to exhibit optimal policy
● Code algorithms related to Markov Decision Process

Prerequisite:
● Linear algebra
● Introduction to Probability and Statistics

Syllabus
● Finite Markov chain: transition matrix, recurrent states, transient states, steady-state, analysis through stochastic matrices, Graph representation: connected graph, Laplacian matrix, Hitting time
● Some examples about Sequential Decision-making process
● Markov Decision Process: the discounted case
o Optimal offline policy, Bellman equation, value function, Q function
o Value Iteration and Policy Iteration algorithms
● Markov Decision Process: the average case (optimal offline policy)
● Constrained Markov Decision Process: optimal offline policy
● Some suboptimal policies: Whittle’s index with threshold policy
● Markov Decision Process without knowing transition probability
o Value and value-cost Q function
o Reinforcement learning: exploration/exploitation tradeoff
o Epsilon-greedy, Boltzmann algorithms
o Deep reinforcement learning: TD algorithm
● Application to Telecommunications (2 Labs)
o Value Iteration
o DRL

Bibliography:
● M. Puterman, “Markov Decision Processes: Discrete Stochastic Dynamic Programming”, 1994.
● E. Altman, “Constrainted Markov Decision Processes”, 1999.
● D. Bertsekas, “Dynamic Programming and Optimal Control”, 1995.
● O. Hernandez-Lerma, “Adaptive Markov Control Processes”, 1989.

UE ML3: Deep Learning

Coordinators:
Hadi Ghauch, Associate Professor, Telecom Paris
Mansoor Yousefi, Associate Professor, Telecom Paris

Volume: 30 hours 3 ects
Hours: Lectures: 27h (4h Paper Presentation session), Seminars: 3h
Assessment: 2 Assignments, 1 Paper Presentation, and Final Project
Language: English

Objectives:
The course offers an in-depth study on the mathematical foundations of deep neural networks (DNNs), which are at the heart of the AI revolution. The first part of the course covers the fundamental aspects of statistical learning and large-scale convex and non-convex optimization methods for modern machine learning tasks. We then focus on learning by a DNN, pose the resulting learning problem as an empirical risk minimization, and discuss state-of-the-art methods of large-scale training. Finally, we describe common DNN architectures/types such as deep convolutional/recurrent neural networks.

Outcomes:
On completion of the course students should be able to:
● Understand fundamental theories on large-scale convex optimization, and non-convex optimization, which underpin deep learning
● Understand state-of-the art training methods being used at the forefront of deep learning research
● Acquire theoretical background (and practical skills) needed to do research in deep learning, or to apply these techniques to their field of expertise

Prerequisite:
● Linear algebra
● Introduction to Convex Optimization
● Introduction to Probability and Statistics

Syllabus
● Fundamentals of deep neural networks (DNNs): mathematical models, feedforward neural networks, derivations of BackPropagation algorithm
● Large-scale training of DNNs: challenges in DNN training, loss surface, state of the art training methods (AdaGrad, RMPProp, ADAM)
● Recurrent neural networks (RNN) for sequence modeling: mathematical models for RNNs, training RNNs (with BPTT), challenges with learning long-term dependencies
● Long-short term memory networks: Motivation, mathematical models, training (with BPTT)
● Convolutional neural networks
● Factor models and manifold learning

Bibliography:
● Goodfellow, Y. Bengio and A. Courville,“Deep Learning”, MIT press, 2016.
● L. Bottou, F. Curtis and J. Norcedal, “Optimization Methods for Large-Scale Machine Learning”, SIAM Rev., 60(2), 223–311, 2018.
● M. Hong, M. Razaviyayn, Z. Q. Luo and J. S. Pang, “A Unified Algorithmic Framework for Block-Structured Optimization Involving Big Data: With applications in machine learning and signal processing”, in IEEE Signal Processing Magazine, vol. 33, no. 1, pp. 57–77, Jan. 2016.

UE COM1: Multi-user Communications

Coordinators:
Philippe Ciblat, Professor, Telecom Paris
Michèle Wigger, Professor, Telecom Paris

Volume: 30 hours 3 ects
Hours: Lectures: 19.5h, Exercises: 4.5h, Labs: 3h
Assessment: 2 Assignments and 1 Final Exam
Language: English

Objectives:
The course considers networks with multiple interfering users. In particular, it discusses techniques for interference mitigation (successive decoding, decoding of non-intended data) and interference avoidance (orthogonal access techniques) and advanced techniques, such as non-orthogonal access, network coding, and resource allocation. Finally, the course also discusses fundamental limits (capacity) of multiuser networks.

Outcomes:
On completion of the course students should be able to:
● Know the standard orthogonal and non-orthogonal multiple access techniques and the most common multi-user receivers
● Characterize and solve resource allocation problems
● Understand some of the most important capacity results for multiple-user networks

Prerequisite:
● Linear algebra
● Introduction to Convex Optimization
● Introduction to Probability and Statistics

Syllabus
● Practical multi-user communication techniques
o Orthogonal multiple access techniques: Time sharing, TDMA, FDMA, OFDMA, CDMA
o Non-orthogonal multiple access techniques : NOMA
o Multi-user decoding techniques: successive interference cancellation (SIC) and parallel interference cancellation (PIC)
● Capacities of networks
o Capacities of Gaussian multi-access channels (MAC) and broadcast channels (BC), MAC-BC Duality
o Cutset upper-bound on the capacity of general networks, Max-flow min-cut theorem
o Linear network coding, Capacity of deterministic multicast networks
● Resource allocation techniques
o Waterfilling
o SNR target
o Interference function and application to Base Transceiver Subsystem (BTS) allocation
o Non-convex case leading to convex one: Geometric Programing, Fractional Programing
o Monotonic programing
o Non-convex case: successive convex approximation, block-coordinate descent
o Biconvex programing
o Relaxation approach (assignment example, second-order cone programing)
o Lab on resource allocation (Yates’ algorithm)

Bibliography:
● D. Tse and P. Viswanath, “Fundamentals of wireless communication”, 2005.
● Goldsmith, “Wireless Communications”, 2005.
● M. Schubert and H. Boche, “Interference Calculus: A General Framework for Interference Management and Network Utility Optimization”, 2012.
● S. Stanczak, M. Wiczanowski, and H. Boche, “Fundamentals of Resource Allocation in Wireless Networks: Theory and Algorithms”, 2008.
● T. Cover and J. Thomas, “Elements of Information Theory”, 2012
● El Gamal and Y.-H. Kim, “Network Information Theory”, 2011.

UE COM2: Advanced Wireless Communications 

Coordinators:
Ghaya Rekaya-Ben Othman, Professor, Telecom Paris
Mireille Sarkiss, Associate Professor, Telecom SudParis

Volume: 45 hours 4.5 ects
Hours: Lectures: 25.5h, Exercises: 9h, Labs: 3h, Seminars: 6h
Assessment: 2 Quizzes and 1 Final Exam
Language: English

Objectives:
The course provides advanced knowledge in a number of transmission techniques and technologies in wireless communications. It covers the fundamentals on MIMO communications ranging from single-user MIMO to multi-user MIMO and cooperative communications, then addresses massive MIMO and mmwave MIMO in contemporary wireless communication standards. Other advanced topics are also viewed to update students with emerging techniques and developments in 5G including: IoT technologies and communications, D2D communications, caching, green communications and various network architectures.

Outcomes:
On completion of the course students should be able to:
● Describe advanced transmission techniques and technologies in wireless communication systems
● Examine major problems of wireless communication channels and propose appropriate solutions
● Design wireless communication systems and investigate further researches in relevant topics

Prerequisite:
● Introduction to Probability and Statistics
● Introduction to Communication Theory

Syllabus
● Basics of single-user Multiple-Input-Multiple-Output (MIMO) communications
o Channel models, outage capacity, ergodic capacity
o Diversity techniques: time, frequency, space and diversity combiners
o Precoding for spatial multiplexing, optimum, linear and nonlinear receivers
o Space-time coding and MIMO decoding
● Advanced MIMO communications
o Multi-user MIMO: interference management, limited feedback and precoding
o Cooperative communications: Relay channels and protocols
o Coordinated Multi-point techniques: Coordinated beamforming and scheduling, joint processing techniques
o Wireless network coding
● Emerging techniques and applications in 5G
o Recent advances in MIMO: massive MIMO and millimeter wave MIMO
o Internet of Things (IoT) networks and Low Power Wide Area Network (LPWAN) technologies (Lora, SigFox, LTE-M, EC-GSM-IoT, NB-IoT)
o Other topics: Caching, Device-to-device (D2D) communications, Massive IoT, Ultra-Reliable Low-Latency Communication (URLLC), Green and energy efficient communications, Mobile Cloud/Edge/Fog computing and C-RAN architectures

Bibliography:
● D. Tse and P. Viswanath, “Fundamentals of wireless communication”, 2005.
● R. W. Heath Jr. and A. Lozano, “Foundations of MIMO Communication”, 2018.
● T.L. Marzetta, E.G. Larsson, H. Yang, H.Q. Ngo, “Fundamentals of Massive MIMO”, 2016.
● E. Bjornsson, J. Hoydis, L. Sanguinetti, “Massive MIMO Networks: Spectral, Energy, and Hardware Efficiency”, 2017.
● Osseiran, J. F. Monserrat and P. Marsch, “5G Mobile and Wireless Communications Technology”, 2016.

UE COM3: Advanced Coding

Coordinators:
Frédéric Lehmann, Professor, Telecom SudParis
Ghaya Rekaya-Ben Othman, Professor, Telecom Paris
Teachers:
Benoît Geller, Professor, ENSTA
Frédéric Lehmann, Professor, Telecom SudParis
Ghaya Rekaya-Ben Othman, Professor, Telecom Paris

Volume: 30 hours 3 ects
Hours: Lectures: 19.5h, Exercises: 9h
Assessment: Final Exam
Language: English

Objectives:
The course provides the state-of-the-art of modern coding theory, along with the corresponding iterative decoding algorithms and an overview of applications in engineering.

Outcomes:
On completion of the course students should be able to:
● Construct a modern (turbo/LDPC/polar) code
● Design a probabilistic decoding algorithm adapted to any modern code construction
● Assess the performances of a modern code for a selected application

Prerequisite:
● Introduction to Probability and Statistics
● Introduction to Information Theory
● Introduction to Communication Theory

Syllabus
● Turbo codes:
o encoding using code concatenation
o iterative (turbo) decoding
o code design and performance analysis
● Low-density parity-check (LDPC) codes:
o construction using sparse graphs
o iterative decoding
o code design and performance analysis
● Polar codes:
o information theory origin
o code construction
o efficient decoding algorithms
● Coding theory for selected applications:
o storage
o cryptography
o codes in standards

Bibliography:
● C. Heegard and S.B. Wicker, “Turbo coding”, Kluwer Academic Publishing, 1999.
● B. Vucetic, “Turbo codes: principles and applications”, Kluwer Academic Publishing, 2000.T. Richardson and R. Urbanke, “Modern coding theory”, Cambridge University Press, 2008.

UE SEC1: Cryptography

Coordinators:
Ghaya Rekaya-Ben Othman, Professor, Telecom Paris
Michèle Wigger, Professor, Telecom Paris

Volume: 30 hours 3 ects
Hours: Lectures: 19.5h, Exercises: 9h
Assessment: 1 Quizz and 1 Final Exam
Language: English

Objectives:
The course provides an introduction to the most important concepts in cryptology such as: Shannon’s cipher, block ciphers, pseudo random number generators, public-key cryptography, different types of attacks, secure hash functions, message authentication, and digital signatures.

Outcomes:
On completion of the course students should be able to:
● Know the most popular encryption systems and known attacks
● Know popular cryptographic concepts such as secure hashing, pseudo random number generators, message authentication, digital signatures

Prerequisite:
● Algebra
● Introduction to Information Theory

Syllabus
● Historic ciphers, Shannon’s cipher and perfect secrecy
● Block ciphers: Data encryption standard (DES), Triple DES, Advanced encryption standard (AES); Types of attacks
● Pseudo random number generators
● Topics in number theory and cryptographic hardness assumptions
● Public-key cryptography: Diffie-Hellman (DH), Rivest-Shamir-Adleman (RSA); Attacks
● Secure Hash Functions: Message digest algorithm (MD5), Secure hash algorithms (SHA); Attacks
● Message authentication
● Digital signatures

Bibliography:
● J. Katz and Y. Lindell, “Introduction to Modern Cryptography », 2007.
● M. Bellare and P. Rogaway, “Introduction to Modern Cryptography », 2005.

UE SEC2: Secure Communications

Coordinators:
Olivier Rioul, Professor, Telecom Paris
Mireille Sarkiss, Associate Professor, Telecom SudParis

Volume: 30 hours 3 ects
Hours: Lectures: 19.5h, Exercises: 6h, Seminars: 3h
Assessment: 2 Quizzes and 1 Final Exam
Language: English

Objectives:
The course describes the information-theoretic basics to secure communications. A first part covers physical layer security techniques exploiting the physical properties of wireless channels. It includes the introduction of wiretap channels, secrecy criteria and measures, coding designs to achieve secrecy, secret key generation and other advanced topics in wireless networks. Then, a second part overviews the problem of side-channel with its main attacks, countermeasures and applications.

Outcomes:
On completion of the course students should be able to:
● Understand and familiarise with the information-theoretic foundations of secure communications
● Analyse the practical physical layer security techniques and describe their limits
● Explain the side-channel analysis attacks and discuss the different countermeasures
● Acquire the current state of research on secure communications and their applications in real scenarios

Prerequisite:
● Introduction to Information Theory
● Introduction to Communication Theory

Syllabus
● Physical Layer Security for wireless communications: Wyner’s Wiretap channel
● Secrecy metrics: weak/strong secrecy, secrecy capacity, secrecy outage probability
● Specific wiretap channels: Gaussian, MIMO, Broadcast Channels (BC) and Multiple-Access Channels (MAC)
● Lattice coding for PHY security and introduction to lattice-based cryptography
● Secure coding schemes: LDPC codes, Polar codes
● Secure network coding
● Secret key generation and secret key agreement
● Other topics: cooperative jamming, secure coding for distributed storage, secure coded caching
● Side-Channel Analysis: physical setup: probing, timing, sounding
o Leakage models in embedded symmetric crypto: monobit, Hamming weight
o Success rate vs. guessing entropy
o Best attacks (Maximum Likelihood (ML), template attacks, Difference of Means (DoM), …) and countermeasures (masking, shuffling, …)
o Asymptotic results: confusion coefficients, exponents
o Information leakage theory and applications

Bibliography:
● M. Bloch and J. Barros, “Physical layer security: From information theory to security engineering », 2011.
● Y. Liang, H. V. Poor and S. Shamai, “Information theoretic security », 2009.
● X. Zhou, L. Song and Y. Zhang, “Physical Layer Security in Wireless Communications », 2013.
● S. Mangard, E. Oswald, and T. Popp, “Power Analysis Attacks », 2007.


Laboratories involved


Information Processing and Communications Laboratory (LTCI), Telecom Paris, Communications and Electronics Department (COMELEC), with 6 professors
SAMOVAR, Telecom SudParis, Communications, Images and Information Processing Department (CITI), with 2 professors
Computer Science and System Engineering Laboratory (U2IS), ENSTA ParisTech, Computer Science and Systems Engineering Department, with 1 professor


Career prospects


The research-oriented master program provides students with a strong academic training on data processing and an in-depth knowledge of communication technologies, machine learning algorithms and information security methods. Students will develop the necessary and multi-disciplinary skills to design concepts and develop solutions in several fields, including Information and Communication Technologies (ICT), biology, healthcare, energy, transportation, and manufacturing.

This unique cross-disciplinary training will allow students to pursue research or development careers in higher education institutions or in industry. Students may also choose to pursue doctoral studies qualifying them later for positions such as researchers and project managers in R&D companies or research fellows and professors in academic institutions.

As a master program of the Institut Polytechnique de Paris, there will be many career opportunities for students with major national and international research centers and higher education institutions.


Institutional partners


MICAS institutions have a strong network of partnerships with leading universities and research institutions around the world. They have established many double degree agreements as well as student and faculty exchange agreements such as Athens, Erasmus+, Magalhães, German-French Academy for the industry of the future, to foster exchanges and provide opportunities for multinational education and research collaborations. MICAS professors have also many national and international collaborations with leading research institutions. A non-exhaustive list of institutional partners is:

● Belgium: Université Catholique de Leuven, Université Catholique de Louvain-la-Neuve
● Brazil: Pontificia Universidade catolica de Rio de Janeiro, Universidade Estadual de Campinas
● China: Beijing University, Shanghai Jiao Tong University, Tsinghua University
● Colombia: Universidad Nacional de Colombia
● Germany: RWTH Aachen University, Technische Universitat Berlin, Technische Universität München
● Greece: Athens University of Economics and Business, Aristote University of Thessaloniki, National Technical University of Athens
● Iran: Sharif University, University of Teheran
● Israël: Technion Israel Institute of Technology
● Korea: Korea Advanced Institute of Science and Technology
● Lebanon: Lebanese University, Université Saint-Esprit de Kaslik, Université Saint Joseph de Beyrouth
● Morocco: Institut national des Postes et Télécommunications de Rabat, Maroc International University of Rabat
● Italy: Politecnico di Milano, Politecnico dii Torino, Università degli Studi di Roma « La Sapienza »
● Russia: Moscow Institute of Physics & Technology, Saint Petersburg State University
● Spain: Universitat Politècnica de Catalunya, Universidad Politécnica de Madrid
● Tunisia : École Nationale d’Ingénieurs de Tunis, École Polytechnique de Tunisie, École Supérieure des Communications de Tunis
● Turkey: Bilkent University
● US: Carnegie Mellon University, Columbia University, New York University, Stanford University, Texas A&M University, University of California Berkeley, University of San Diego
● Sweden: KTH Royal Institute of Technology
● Switzerland: École Polytechnique Fédérale de Lausanne, ETH Zürich


Industrial partners


MICAS institutions and professors collaborate actively with many industrial partners through Ph.D. theses projects and research contracts. These partners often propose internships to students. A non-exhaustive list of partners is:

● Thalès
● Nokia Bell Labs
● Huawei Technologies
● Orange Labs
● EDF
● Ebay
● D.E. Shaw
● Mitsubishi Electric
● Merck & Co.
● SagemCom
● Uber
● CEA
● INRIA


Admissions


Application guidelines for a master’s program at IP Paris

Academic Prerequisites
The applications are eligible for:
● Students registered in the first-year master (M1) in France
● Foreign students with a four-year Bachelor degree from an approved university or institution
● European Erasmus students already registered in their home Master institution
● Students with an M.Sc. degree or Engineering degree (french or foreign)
Irrespective of their domains, students are expected to be mathematically minded and have a solid background in mathematics.

Language prerequisites
● Good knowledge of English language

Application timeline
Deadlines for the Master application sessions are as follows:
– First session: February 28, 2020
– Second session: April 30, 2020
– Third Session (optional): June 30, 2020 (only if there are availabilities remaining after the 2 first sessions)
Applications not finalized for a session will automatically be carried over to the next session.

You shall receive an answer 2 months after the application deadline of the session.


Tuition fees


International Master: EU students : 4250 euros / Non-EU students: 6250 euros


Contact


Sarkiss Mireille
Telecom SudParis

Tchamkerten Aslan
Telecom Paris

Email