Wireless Communications

Algorithms and Building Blocks for Communications

MIMO Performance Indicators for Industrial Design Use

Master Thesis

Description: The objective of this project is to compare different ways to estimate the Multiple-Input Multiple-Output (MIMO) performance of an antenna system from the point of view of an industrial design usage. MIMO systems are more and more becoming an integral part of modern communication systems. For example, they are already included in LTE mobile communication standard, in 802.11n and later versions of WiFi communication standard, as well as in the future communication standards, as for example 5G mobile networks. Currently, different metrics are used to assess the MIMO gain potential of antenna systems. They are ranging from simple factors like isolation between the radiators and envelope correlations computed using S-parameters of an antenna to the correlation computed using antenna 3D patterns and channel spatial power distribution knowledge. The aim of this activity is to review all possible solutions and establish guidelines for industrial antenna development and experimental testing. This includes also a numerical and experimental evaluation of existing HUBER+SUHNER MIMO antenna designs.

Company details: The global Swiss company HUBER+SUHNER develops and manufactures components and system solutions for electrical and optical transportation of data and energy. The company serves customers in the Communication, Transportation and Industrial markets with cables, connectors, cable systems, antennas and other passive components relying on its expertise in radio frequency, fibre optics and low frequency technologies.

Within our Antennas Research & Development team we offer the opportunity to complete a master thesis project in collaboration with Ecole Polytechnique Fédérale de Lausanne (EPFL). The H+S R&D team Antennas supports a broad range of products for the above mentioned three markets with applications to WLAN / WiFi, Distributed Antenna System (DAS), GPS, Vehicles or also combinations with MIMO or other innovative solutions up to millimetre wave frequencies.

Kind of work: The work consists of theoretical work, numerical algorithms development and testing as well as experimental antenna tests and measurements. Different RF design tools may be used like CST or MATLAB as well as others. Additionally, the candidate will obtain a hands-on experience with RF measurements and testing. By doing this, the candidate will cover and become familiar with many antenna design aspects joining the HUBER+SUHNER Antenna development team.

Supervisor: contact Prof. Andreas Burg (andreas.burg@epfl.ch)

Date added: 04-02-15

Efficient Characterisation of Wireless Channels via Compressive Sensing


Semester Project

Description: Reliable models of the wireless channel are an integral part of the design and analysis tools used to predict end-to-end performance of modern telecommunications systems. These tools can be used by system planners to investigate and evaluate the impact of different deployment strategies, antennas, modulation schemes, and other related parameters. However, developing reliable models for the wireless channel often requires extensive measurement campaigns, which are time-consuming and expensive. In particular, to correctly account for multi-path fading, measurements of the channel must be taken at many different points in space, e.g., over a spatial grid using a positioner. 

The goal of this project is to use compressive sensing to significantly reduce the number of measurement points that are required to develop reliable models of the wireless channel. Compressive sensing is a recent signal processing technique that can very efficiently reconstruct an under-sampled signal by exploiting sparsity in one or more of the signal domains. As the spatial variation in signal strength caused by multi-path fading is a sparse-signal, the fading pattern can be reconstructed from a small number of measurements by applying compressive sensing. The compressive sensing algorithms will be developed and verified from actual measurements of the channel, using the TCL wireless testbed and 2D positioner. 

Prerequisites: Basic background in signal processing and wireless channel characterisation. Some knowledge of Labview and MATLAB is an advantage.

Project Type: 30% Labview, 30% Measurements, 40% MATLAB

Supervisor: Andrew Austin

Date added: 04-02-15

Using APSK: LDPC or Polar Codes?

Description: Many systems accept a performance penalty by using square-shaped constellations, e.g., QAM. Using a round-shaped constellation like APSK can result in a 1 dB gain in mutual information compared to QAM. In this project, we would like you to investigate which modern channel code performs best: low-density parity-check (LDPC) codes or polar codes?

Project goals:

  • Conduct a rigorous study
  • Present analytical results
  • Confirm with simulation

Areas: Communication, modulation, channel coding, MATLAB

Supervisors: Pascal Giard, Matthieu Cotting, and Orion Afisiadis

​​​​​​​Date added: 18.12.2017

Communications System Design and Prototyping

Power efficient, high throughput, flexible Polar decoder

Master Thesis or Semester Project



Description: Polar codes are a new and promising type of error correcting codes. Invented in 2009, they have quickly advanced to one of the hottest topics in coding theory and they are a likely candidate for the next generation of many communication standards. The beauty of these codes lies in the fact that they can provably asymptotically achieve the channel capacity for various types of communications channels. At the same time, decoding of Polar codes is possible using a successive cancellation scheme that has a computational complexity that is low enough for practical applications.

The implementation of decoders for Polar codes has not received much attention yet. The feasibility of high throughput and power efficient implementations of such decoders is an important prerequisite for the possible adaptation of Polar codes by future communications standards. The first ASIC implementation of a Polar decoder was recently completed in our lab. The goal of this project is to investigate throughput and power efficiency enhancing optimizations of the existing design.

To this end, the candidate will first study and refine a MATLAB model of the decoder, along with a brief review of the required theoretical background. Furthermore, various improvements of the decoder will be considered and implemented in VHDL in order to assess their effect on throughput and power consumption. Currently, the decoder can only decode a specific rate-1/2 code of length 1024 bits. So, another possible improvement is to modify the decoder in order to make it more flexible.

Prior knowledge: Interest in communications & digital signal processing, VLSI Design, VHDL.

Kind of work: 20% MATLAB simulation and theoretical background, 40% Architecture Design, 40% VHDL.

Supervisor: Alexios Balatsoukas-Stimming



Full-Duplex LoRa-inspired ACK Scheme

Description: ALOHA leads to collisions and retransmissions. LoRa requires as little downlink as possible, has stringent duty-cycle restrictions. We propose to tackle these issues using a full-duplex LoRa-inspired end-node, i.e., to transmit on the uplink, incrementally adding redundancy, until gateway is able to decode and send an ACK.

Project goals:

  • Examine possible rateless coding schemes
  • Build a real-time full-duplex link with this novel hybrid ARQ scheme

Areas: Wireless communications, channel coding, MATLAB, LabVIEW

Supervisors: Orion Afisiadis, Matthieu Cotting, and Pascal Giard

Date added: 18.12.2017

LoRa I/Q Framer for GNU Radio

Description: We have a reference LoRa transmitter built with LabVIEW. It supports only one code rate and one spreading factor, but has been verified and works well. We would like to build a transmitter using the GNU Radio software-defined radio (SDR) framework. This project consists in implementing part of that transmitter, namely from the I/Q Framer to the communication with a USRP SDR. The I/Q Framer shall take digital symbols as an input, convert them to I/Q symbols and prepend the preamble and synchronization sequence. The goals include both the implementation and building a working prototype.

Areas: Communication, signal processing, SDR, C/C++, and Python

Supervisors: Matthieu Cotting, Pascal Giard, and Orion Afisiadis

Date added: 18.12.2017

LoRa Packet Generator and

Symbol Mapper for GNU Radio

Description: We have a reference LoRa transmitter built with LabVIEW. It supports only one code rate and one spreading factor, but has been verified and works well. We would like to build a transmitter using the GNU Radio software-defined radio (SDR) framework. This project consists in implementing part of that transmitter, namely the packet generator and the symbol mapper. The packet generator takes a codeword (data) as an input and adds the header, the header cyclic redundancy check (CRC), and the data CRC. The symbol mapper maps the binary stream to digital symbols.

While the latter mapping is well defined, the header format and CRC parameters need to be reversed engineered with some hints from the patent as well as experiments using commercial off the shelf LoRa-compliant devboards.

Project Goals:
• Reverse engineer the header format
• Reverse engineer the CRC parameters
• Create a reference implementation in software or hardware (e.g., MATLAB, Python, C, VHDL)

Areas: Communication, signal processing, reverse engineering, and C or VHDL

Supervisors: Orion Afisiadis, Matthieu Cotting, and Pascal Giard

Date added: 18.12.2017

Part of the LoRa Digital Transmitter Chain

From User Data to Codeword

Description: We have reference LoRa transmitters built with MATLAB and LabVIEW. We would like to build transmitters in both software and hardware. This project consists in implementing part of that transmitter, namely part of the digital chain.


  • Hamming encoder with 4 code rates
  • Flexible interleaver
  • Whitening
  • Gray mapping

Each component have to accommodate for 6 possible spreading factors (7 to 12), in addition to that, the Hamming encoder needs to support multiple code rates. Note that we have a reference implementation in both MATLAB and LabVIEW. The LabVIEW implementation supports only one code rate and spreading factor.

Project goals:

  • 1 Student shall implement this chain in software (C language)
  • 1 Student shall implement this chain in hardware (VHDL language)

Areas: Communication, signal processing, and C or VHDL

Supervisors: Reza Ghanaatian, Pascal Giard, and Matthieu Cotting

Date added: 18.12.2017

Tools for Communications

Software tool for performance assessment of IQ digital modulations

Concerned sections: SEL, SSC

Project Type: semester master (projet d’électricité II 11 ECTS)

Description:  This project aims to build a tool intended to investigate the performances of digital modulations based on IQ technique. Thus, it is well known that using two carriers, one in-phase (I) and one in-quadrature (Q) one can produce all common QAM or PSK digital modulations. The IQ tool will be programmed in the VEE simulation environment (similar to Labview) and will include a complete simulation of the transmission system: matching filter, modulator, the channel as a band-pass filter, demodulator, and receiver filter. IQ tool will also include an eye-pattern analyzer and a BER estimator. Some of these blocks are already available. 

Using the IQ tool, several studies will be performed regarding the influence of channel limitations (bandwidth, linear and nonlinear distortions, etc.) on the performance of QAM and PSK modulations.

Kind of work: 60% VEE programming, 30% performance assessment, 10% report and documentation.

Achievements: deep knowledge in digital IQ modulation, experience in digital signal processing, good preparation for master projects in electronics and telecom systems.

Prior knowledge: IQ technique, QAM and PSK modulations. 

Supervisor: Christian Gaumier


Graph recognition algorithms

Concerned sections: INF, SSC, SEL

Project Type: semester master

Description:  This project aims to create and implement some graph recognition algorithms. Various practical applications, such as automatic circuit analysis, require the detection of some basic graph structures within big complex graphs. Thus, the detection can include either topological aspects (vertices and edges) or edge-related attributes. The matching degree can be described by some distance metrics.

The goal of the project is to implement various graph recognition algorithms and to evaluate their performances on several case studies. The recommended programming environment is ATAG which allows the use of scripts written in Groovy (based on Java language). 

Kind of work: 70% programming, 20% performance assessment, 10% report and documentation.

Achievements: experience in algorithmic graph theory, algorithm development, ATAG environment, programming in Groovy. Valuable basis for a master project in algorithm development and graph related algorithms.

Prior knowledge: Basic algorithmic graph theory, Java programming language, potentially Groovy.

Supervisor: Christian Gaumier