Search results

Create the page "In Progress" on this wiki! See also the search results found.

• Scalable Heterogeneous L1 Memory Interconnect for Smart Accelerator Coupling in Ultra-Low Power Multicores
  ...level semantically rich information out of raw data is deep learning, and in particular deep convolutional neural networks (CNNs). The task of inference ...tand-alone ASICs for aggressively quantized CNNs (''YodaNN'' [Andri2017]), in the Ergo project we want to design a PULP-based entire computation cluster
  7 KB (961 words) - 21:21, 29 January 2019
• Timing Channel Mitigations for RISC-V Cores
  ...axis). The effect of timing mitigation techniques can be clearly observed in the right matrix, where almost no correlation can be observed anymore. Imag ...are essentially a combination of speculative execution techniques present in today's processors, combined with shared hardware state and timing channels
  6 KB (915 words) - 18:16, 20 May 2020
• Developing a small portable neutron detector for detecting smuggled nuclear material
  ...s. The main technological challenge of detecting fast neutrons is doing so in a way that distinguishes such detected particles very reliably from detecte ...urity applications. This is the purpose of this project. There is interest in this project for example by Labor Spiez (Swiss ABC-protection agency). Due
  5 KB (623 words) - 10:32, 5 November 2019
• Design of an on-field adaptable pulse-processing unit for semicondutor radiation detectors
  Semiconductor devices can be exploited as radiation detectors in a variety of applications with their combined use with spectrometry systems ...shortcomings of a basic implementations become apparent when this is used in combination with a non-ideal detector.
  4 KB (506 words) - 18:26, 5 November 2019
• ML based Quantitative Movement Analysis on a Portable IoT Camera (1-2S/B)
  ...ency and lead to injuries. In clinical biomechanics, understanding changes in muscle and tendon properties is vital for creating effective treatment plan ...em should be easy and quick to use (e.g., time is key for testing patients in clinics) and inexpensive compared to traditional necessary hardware.
  6 KB (735 words) - 12:12, 23 July 2023
• Floating-Point Divide & Square Root Unit for Transprecision
  ...out all computations at the maximum available precision - double precision in the case of floating-point (FP)workloads. This happens regardless of the ac ...where the precision of the entire system is reduced - often incurring loss in result quality - transprecision computing entails dynamically providing the
  8 KB (1,135 words) - 17:09, 29 July 2020
• Design and implementation of the front-end for a portable ionizing radiation detector
  ...emory), or a catastrophic failure of the device (e.g. Single Event Burnout in high power devices). Since the interaction of ionizing particles with matte ...evice reliability is becoming a primary challenge requiring new approaches in terms of device physics, device modelling, physics of failure, thermal mana
  5 KB (615 words) - 12:23, 9 May 2019
• Channel Estimation for 5G Cellular IoT and Fast Fading Channels
  ...g channels is thus needed to improve the performance of the current system in such scenarios. ...ising complexity-performance trade-off you will implement and compare them in our Matlab framework. You will then select one algorithm, and optimize and
  3 KB (415 words) - 18:54, 29 October 2020
• Subject specific embeddings for transfer learning in brain-computer interfaces
  ...akes it particularly challenging, however, is its susceptibility to errors in the recognition of human intentions. The underlying reason is the low signa * project report (in digital form; a hard copy also welcome, but not necessary)
  4 KB (585 words) - 11:12, 21 July 2020
• FPGA Testbed Implementation for Bluetooth Indoor Positioning
  ...LE packets and estimating their received signal power was developed at IIS in previous projects. ...power. The previously developed digital baseband receiver block integrated in a processor system shall be used as a starting point. As a first step you w
  3 KB (431 words) - 21:47, 18 November 2019
• Experimental Validation of Impact Ionization Models for TCAD Simulation by a Novel Characterization Technique
  Modeling and simulation of impact ionization in semiconductors is fundamental to the design and the robustness of electroni ...se to the breakdown. However, this approach is not very accurate, because, in real devices, the breakdown first occurs at small isolated spots or lines (
  5 KB (665 words) - 10:30, 5 June 2019
• Design and Exploitation of a Test-Bench for Non-Destructive Characterization of the Susceptibility of Silicon Carbide (SiC) Power Devices to Cosmic Radiation
  ...fferent types of electromagnetic and/or nuclear interactions. This results in the deposition of energy (charge) along the ionization track that may be su ...ated conditions, i.e. by irradiating such devices with light or heavy ions in particle accelerators. However, these experiments require special facilitie
  4 KB (493 words) - 16:31, 5 June 2019
• PREM Intervals and Loop Tiling
  ...only offer a major increase in performance, but also promises an increase in energy efficiency due to the use of more efficient accelerators. However, c ...program when the system is not permitting memory accesses from the program in question, the memory phase is tasked with copying all data needed for compu
  4 KB (531 words) - 18:00, 19 June 2019
• Compiler Profiling and Optimizing
  ...only offer a major increase in performance, but also promises an increase in energy efficiency due to the use of more efficient accelerators. However, c ...program when the system is not permitting memory accesses from the program in question, the memory phase is tasked with copying all data needed for compu
  3 KB (466 words) - 18:20, 19 June 2019
• Extending the RISCV backend of LLVM to support PULP Extensions
  ...V ISA has been rapidly increasing in popularity, in academia, but now also in industry. The LLVM compiler infrastructure has been recently extended to co : Interest in compiler techniques
  2 KB (300 words) - 18:27, 19 June 2019
• Next Generation Synchronization Signals
  ...oss-correlations, which is one of the most computationally expensive tasks in an IoT modem. Recently, a new kind of synchronization waveforms has been pr ...oject you will look at the proposal in detail and evaluate its performance in comparison to state-of-the-art solutions. For this you can use our existing
  3 KB (358 words) - 10:51, 5 April 2022
• Energy Efficient Smart Devices For Construction Building Maintenance Hilti Collaboration
  ...n IIS and Hilti that includes the desing of new hardware-software devices. In particular, machine learing, energy harvesting, bluetooth communication and ...nto existing tools and store the real time data while the tool is running. In the range of 100Hz and 1 to 8 IO ports with differing signal types. Possibl
  5 KB (645 words) - 15:41, 10 November 2020
• Wearable Smart Camera With Deep Learning Algorithms For Automatic Detecion
  ...rator is planned to be achieved by the student. This work can also be done in collaboration with armasuisse. [[Category:In progress]]
  4 KB (519 words) - 15:41, 10 November 2020
• Adversarial Attacks Against Deep Neural Networks In Wearable Cameras
  ...stringer requirements of low power consumption. This work can also be done in collaboration with armasuisse. [[Category:In progress]]
  4 KB (509 words) - 15:41, 10 November 2020
• Neuromorphic Intelligence In An Embedded System in Collaboration with AiCTX
  ...ological edge and IP portfolio that comes from over 20 years of experience in mixed-signal neural processor design, advanced neural routing architectures ...from 2019 IIS and aiCTX are providing several semester and master projects in the field of neuromorphic intelligence using their processor to build a who
  5 KB (692 words) - 15:45, 10 November 2020
• High-throughput Embedded System For Neurotechnology in collaboration with INI
  ...and the main goal is to build a whole working prototype tha will be tested in-field. [[Category:In progress]]
  4 KB (526 words) - 15:48, 10 November 2020
• Design Of A Biomarker Assay Based On Responsive Magnetic Nanoparticles
  30% Implementation (including construction and testing in the lab) 40% Iteration in design and implementation
  4 KB (500 words) - 15:48, 10 November 2020
• Manycore System on FPGA (M/S/G)
  In a quest for high-performance computing systems, few architectural models re ...stants’ schedule. These meetings will be used to evaluate the status and progress of the project. Besides these regular meetings, additional meetings can be
  8 KB (1,319 words) - 10:41, 6 July 2021
• Hardware Acceleration
  ...t. Such accelerators are not limited to the high-performance sector alone. In low power computing, they allow complex tasks such as computer vision or cr ...monolithic accelerators dedicated to specific tasks are usually unbeatable in terms of throughput and energy efficiency, they also have drawbacks: they a
  7 KB (917 words) - 17:04, 24 November 2023
• Signal Acquisition and Clock Offset Compensation for High-Rate Pulse UWB PHYs
  [[File:3db_logo.png|600px|thumb|M.S. project in collaboration with 3db Access on UWB-based localization]] ...enting Low-Rate Pulse (LRP) UWB technology. By introducing this technology in a classical key fob, VW demonstrates the feasibility of an ultra-low power,
  5 KB (584 words) - 12:09, 29 October 2020
• Finite element modeling of electrochemical random access memory
  ...research teams in Zurich and the IBM T.J. Watson Research Center in Yorktown Heights, NY, USA. ...ercalation of an ion into a host material, which causes a change in electrical resistance. Compared to other NVMs such as phase-change
  4 KB (515 words) - 17:06, 16 September 2021
• Phase-change memory devices for emerging computing paradigms
  ...rning inference hardware, where the neural network layers would be encoded in crossbar arrays of memory devices [2]. However, there are several challenge : [1] A. Sebastian et al. Memory devices and applications for in-memory computing. Nature Nanotechnology (2020). https://doi.org/10.1038/s4
  5 KB (640 words) - 14:13, 23 June 2021
• Electrothermal characterization of van der Waals Heterostructures with a partial overlap
  simulations in vibrational, optical, and electronic properties and transport properties of monolayers are stacked on top of each other only in their central part.
  3 KB (372 words) - 17:06, 16 September 2021
• Implementation of Computationally Efficient Scattering Mechanisms for Periodic Devices and 2D Materials
  are periodic in at least one of the three spacial dimensions. 2D materials have seen a surge of interest in recent years for their advantageous electronic properties. For
  2 KB (327 words) - 17:06, 16 September 2021
• Ferroelectric Memristors for Artificial Neural Networks (IBM-Zurich)
  At the Neuromorphic Devices and Systems laboratory in IBM Research, we have an opening for a Master research project: ...ld-effects and electron tunneling, to understand the conduction mechanisms in the characterized devices.
  2 KB (320 words) - 12:21, 23 June 2021
• Towards global Brain-Computer Interfaces
  ...akes it particularly challenging, however, is its susceptibility to errors in the recognition of human intentions, especially during motor imagery (MI). ...ed that we can find a global model for 9 subjects on a 4-class MI dataset. In this project, you apply hyperdimensional superposition on a much larger dat
  5 KB (672 words) - 09:20, 16 September 2021
• EECIS
  category = In progress category = In progress
  740 bytes (70 words) - 15:18, 25 September 2019
• Energy Efficient Circuits and IoT Systems Group
  category = In progress
  736 bytes (72 words) - 15:19, 19 December 2022
• Toward hyperdimensional active perception: learning compressed sensorimotor control by demonstration
  ...as well as universal computing, e.g., emulating finite state automata [2]. In this project, your goal would be to improve algorithms, and implementations for robotics applications, especially with the dataset studies in [3].
  2 KB (308 words) - 20:12, 9 February 2020
• TCNs vs. LSTMs for Embedded Platforms
  ...t task. Another class of neural networks which achieved impressive results in visual tasks is the Convolutional Neural Networks (CNNs), a variation of ...e student will proceed with embedded implementation and system integration in order to demonstrate a real-life application using sensor acquisition and m
  2 KB (317 words) - 11:10, 21 July 2020
• An Energy Efficient Brain-Computer Interface using Mr.Wolf
  ...akes it particularly challenging, however, is its susceptibility to errors in the recognition of human intentions. To cope with high variability of EEG d ...While being successful in image classification, CNNs are gaining attention in MI-BCIs as well [2]. One of the most popular and compact CNNs is EEGNet [3]
  3 KB (372 words) - 11:09, 21 July 2020
• BirdGuard
  ...or this application. The algorithm shall be validated on a camera/PC setup in the field and later optimized for low power execution on the ML chip to ena ...d distance). The results shall be implemented in a prototype and validated in a field pre-study.
  5 KB (647 words) - 08:32, 23 February 2023
• Characterization of the susceptibility to cosmic radiation of wide bandgap power devices by radioactive sources and at cryogenic temperatures
  ...erize the internal electric field of silicon power devices has been proven in [https://doi.org/10.1016/j.microrel.2019.06.035 I]. This solution allows to ...nt at ambient temperature and the low charge deposition of gamma radiation in semiconductor devices, the charge deposition induced signal may be complete
  4 KB (473 words) - 18:33, 5 November 2019
• Pressure and acoustic Smart Sensors Network for Wind Turbines Monitoring
  In this project, we aim to develop a first-ever MEMS-based surface pressure an ...of the system will be performed from the students as an important activity in order to evaluate power consumption, reliability, functionality, classifica
  5 KB (737 words) - 15:48, 10 November 2020
• Semi-Custom Digital VLSI for Processing-in-Memory
  ...reak through the memory wall. The concept of PIM has been largely explored in many recent papers, but most of these works have focused on using some kind Recently, we have proposed PPAC (Parallel Processor in Associative Content-Addressable Memory) [1], a PIM architecture that is abl
  7 KB (882 words) - 14:33, 28 July 2021
• Ibex: FPGA Optimizations
  ...logna under the name Zero-riscy [1], and has been taped out multiple times in a mix of academic and industry projects. Since December 2018, Ibex is activ ...SIC designs and is optimized for low area footprint and energy consumption in ASIC technologies. On FPGAs, Ibex currently has a higher resource utilizati
  5 KB (666 words) - 09:45, 28 August 2020
• Hyper-Dimensional Computing Based Predictive Maintenance
  the remaining useful lifetime. This allows replacing machine components in a traditional machine learning techniques were data is represented in the form of
  5 KB (759 words) - 09:18, 16 September 2021
• Ibex: Bit-Manipulation Extension
  ...logna under the name Zero-riscy [1], and has been taped out multiple times in a mix of academic and industry projects. Since December 2018, Ibex is activ ...However, the simplicity currently comes at the price of lower performance in terms of cycles per instruction compared to bigger cores such as RI5CY [1].
  4 KB (639 words) - 09:45, 28 August 2020
• Contrastive Learning for Self-supervised Clustering of iEEG Data for Epileptic Patients
  ...ging for certain types of data that are highly complex or noisy, resulting in poor quality human annotations at best. A promising alternative is to train the models in a self-supervised way using contrastive learning [1], which was able to lea
  5 KB (650 words) - 20:01, 10 March 2024
• Exploring features and algorithms for ultra-low-power closed-loop systems for epilepsy control
  ...an abnormal signal that allows tailoring a therapy that is to be delivered in a closed-loop manner. ...s verified in real time. This technique is used in the control of epilepsy in closed-loop fashion, for example. Still, improving the efficiency of the de
  3 KB (381 words) - 11:22, 21 July 2020
• Deep neural networks for seizure detection
  ...ze both the energy efficiency for years-operating devices and the accuracy in the epilepsy detection. In terms of time-series analysis, a big variety of deep-learning approaches ar
  6 KB (780 words) - 20:02, 10 March 2024
• Mapping Networks on Reconfigurable Binary Engine Accelerator
  ...flexible and power-scalable way. In this project, you will make us of our in-house developed frameworks NEMO [2] (or Quantlab [3]) and DORY[4,5] to map ...ws the RBE to perform convolutions with configurable arithmetic precisions in a flexible and power-scalable way. BBQ can be applied on both complete NNs
  6 KB (814 words) - 09:55, 8 March 2023
• Accurate deep learning inference using computational memory
  ...ces are used to perform certain computational tasks within the memory unit in a non-von Neumann manner. ...rning inference hardware, where the neural network layers would be encoded in crossbar arrays of memory devices [2]. However, there are several challenge
  5 KB (628 words) - 12:51, 17 April 2020
• ASIC Implementation of Jammer Mitigation
  ...the legitimate user equipments. The signal processing for this takes place in a custom ASIC.]] ...be practically viable, such methods will ultimately have to be implemented in hardware (using FPGAs or, more likely, ASICs), since sofware-based signal p
  5 KB (662 words) - 13:31, 10 May 2023
• Real-Time Motor-Imagery Classification Using Neuromorphic Processor
  ...ation for severely paralyzed users, control of a wheelchair, or assistance in stroke rehabilitation. ...y and a couple of classes motor-imagery and often they are not implemented in a neuromorphic processor, and none of them are presenting a whole system fr
  6 KB (815 words) - 20:02, 10 March 2024
• Andrea Cossettini
  ...[https://www.infineon.com/cms/austria/en/ Infineon Technologies] (Austria, in 2014-​​​2015). During his PhD, he worked on [https://air.uniud.it/ret '''Most of the projects evolve very fast. If you are interested in the research areas mentioned above, send me a message to discuss up-to-date
  3 KB (406 words) - 17:17, 3 May 2024
• BCI-controlled Drone
  ...akes it particularly challenging, however, is its susceptibility to errors in the recognition of human intentions, especially during motor imagery (MI). In this project, the student works on the acquisition and processing of electr
  5 KB (669 words) - 15:27, 23 October 2023
• Compression of iEEG Data
  * project report (in digital form; a hard copy also welcome, but not necessary) ...aud, et al., Unsupervised Learning of Spatiotemporal Interictal Discharges in Focal Epilepsy, 2018
  5 KB (636 words) - 20:01, 10 March 2024
• Data Augmentation Techniques in Biosignal Classification
  ...student explores data augmentation methods [1][2] to improve the learning in biosignal classification tasks. Applications range from ECG to EEG [3] clas * project report (in digital form; a hard copy also welcome, but not necessary)
  4 KB (585 words) - 19:59, 10 March 2024
• Classification of Evoked Local-Field Potentials in Rat Barrel Cortex using Hyper-dimensional Computing
  ...g neural circuits and, in perspective, to operate neuroprosthetic devices. In this context, artificial intelligence combined with low-power embedded devi ...ion amplitudes. The aim of the project is to assess the performance of HDC in classifying different external stimulus applied to the animal.
  5 KB (619 words) - 19:58, 10 March 2024
• Improved State Estimation on PULP-based Nano-UAVs
  ...ger in many practical applications such as aerial inspection or assistance in natural disaster areas, to name a few. In this context, the parallel ultra-low-power (PULP) platform[4] developed her
  8 KB (1,117 words) - 22:17, 26 January 2022
• Deep Learning-based Global Local Planner for Autonomous Nano-drones
  ...ger in many practical applications such as aerial inspection or assistance in natural disaster areas, to name a few. A common approach in the state-of-the-art (SoA) of autonomous driving systems is represented by
  4 KB (571 words) - 12:11, 27 January 2022
• RISC-V based Implementation of Secure Ranging According to IEEE 802.15.4z UWB
  ...Events in IoT and Embedded]. After Apple quietly introduced the U1 chipset in its high-end iPhone 11, Volkswagen, the largest automaker, also unveiled fi ...ng the Low-Rate Pulse (LRP) UWB technology. By introducing this technology in a classical key fob, VW demonstrates the feasibility of an ultra-low-power,
  8 KB (1,139 words) - 19:10, 29 January 2021
• An all Standard-Cell Based Energy Efficient HW Accelerator for DSP and Deep Learning Applications
  In the past couple of years the Integrated Systems Laboratory (IIS) developed ...ifferences in minimal operating voltage can thus lead to large differences in power consumption and to a certain degree also energy efficiency.
  7 KB (1,032 words) - 15:31, 16 November 2020
• Memory Augmented Neural Networks in Brain-Computer Interfaces
  ...sage of pure convolutional neural network (CNN) architectures very limited in this area, and less practical for multi-users deployment. A viable option a ...at the beginning of a new session to mitigate high inter-session variance in EEG.
  5 KB (662 words) - 21:31, 9 January 2022
• Combining Spiking Neural Networks with Hyperdimensional Computing for Autonomous Navigation
  In the last 10 years, Artificial Neural Networks (ANN) revolutionized many sci popular approach which allowed to reach state-of-the-art accuracies in many different machine
  5 KB (707 words) - 09:23, 16 September 2021
• High Performance SoCs
  ...data increases. Despite continuous advances, memory lags behind computing in scaling, and many data-driven problems today are memory-bound. ...n streams of cores may compete for valuable memory bandwidth and interfere in a destructive way.
  11 KB (1,337 words) - 10:54, 25 January 2024
• A Flexible Peripheral System for High-Performance Systems on Chip (M)
  Although our idea works well in theory, we have not implemented any peripherals or a demonstrator system ye ...l system for high-performance and energy-efficient SoCs and demonstrate it in a compact RISC-V-based microcontroller on real silicon.
  11 KB (1,675 words) - 15:40, 15 March 2021
• Tbenz
  ===Projects In Progress=== category = In progress
  362 bytes (31 words) - 16:48, 29 July 2020
• A Snitch-based Compute Accelerator for HERO (M/1-2S)
  ...ts default configuration [<nowiki/>[[#ref-snitch|4]]] |The Snitch cluster in its default configuration [<nowiki/>[[#ref-snitch|4]]] ]] ..., peripherals, and a tightly-coupled memory form a Snitch cluster as shown in [[#fig:snitch_bd|2]].
  11 KB (1,617 words) - 23:59, 6 February 2021
• Peak-to-average power Reduction
  ...plementation, or purely VLSI implementation based. The tools to be learned in this project are numerical (convex) optimization and deep unfolding, a rece : For students interested in the theory: Communication Systems (''recommended'')
  4 KB (513 words) - 14:16, 24 November 2021
• Hardware Constrained Neural Architechture Search
  ...nsumption, and latency. This results in the search space for manual design in being combinatorially large. A method of tackling the problem of manual des ...akes it particularly challenging, however, is its susceptibility to errors in the recognition of human intentions, especially during motor imagery (MI).
  6 KB (784 words) - 21:34, 9 January 2022
• Visualization of Neural Architecture Search Spaces
  ...nsumption, and latency. This results in the search space for manual design in being combinatorially large. A method of tackling the problem of manual des In this project, the student will work with a tool engineered at the lab for t
  5 KB (760 words) - 01:39, 10 January 2022
• Knowledge Distillation for Embedded Machine Learning
  ...hows a lot of potential for deploying models to embedded devices when used in conjunction with well-established quantization techniques. The goal of this The main goals (not all have to be met in a single semester project) of the project are:
  5 KB (670 words) - 15:14, 4 August 2022
• Energy Efficient SoCs
  As part of our research activities we regularly tape-out PULP based SoC in various sizes, from smaller student tape-outs incorporating novel hardware ===Projects In Progress===
  3 KB (339 words) - 15:59, 1 November 2023
• VLSI Implementation of a 5G Ciphering Accelerator
  ...s and self-driving cars demand for reliable and secure communication links in order to avoid dangerous malfunctions or external attacks. 5G cellular netw ...fficient ciphering architecture meeting the low-latency demands for URLLC. In a second part of the project, the accelerator architecture will be ported t
  3 KB (404 words) - 10:05, 9 February 2021
• Nils Wistoff
  ===Projects In Progress=== category = In progress
  578 bytes (56 words) - 18:59, 30 October 2020
• Coherence-Capable Write-Back L1 Data Cache for Ariane (M)
  [[Category:In progress]] == Status: In Progress ==
  3 KB (395 words) - 16:32, 15 November 2022
• Event-Driven Vision on an embedded platform
  ...occur and entering an idle mode when there are no events. This can be used in the context of machine learning to only trigger network inference when even ...ognition on embedded platforms. Depending on the student’s interests and progress, a full working sensor node including Bluetooth communication and deploymen
  3 KB (449 words) - 08:41, 17 February 2021
• Physical Implementation of MemPool, PULP's Manycore System (1M/1-2S)
  In a quest for high-performance computing systems, few architectural models re Our example system, in GlobalFoundries 22nm 22FDX technology, runs at 500MHz, and the L1 memory ca
  8 KB (1,196 words) - 10:41, 6 July 2021
• Feature Extraction and Architecture Clustering for Keyword Spotting (1S)
  ...be largely influenced by this choice, the aim of this project is to do an in-depth evaluation of feature extraction methods for KWS and their pairing wi ...xtraction, defining the former as the techniques applied to alter the data in order to emphasize or remove certain characteristics. To better understand
  13 KB (1,815 words) - 13:31, 15 February 2024
• Self-Supervised User Positioning in Cell-Free Massive MIMO Systems
  ...NSS information and satellite-based localization is unreliable indoors and in dense rural scenarios. As a consequence, current research focuses on develo ...uted APs has the potential to significantly increase localization accuracy in indoor and rural scenarios, while avoiding the need of labeled training dat
  8 KB (931 words) - 17:27, 23 November 2021
• Matteo Perotti
  I am interested in everything is hardware, from highly efficient low-power implementations to I mainly work on RISC-​V architectures, and I am secretly in love with the SW-​side of the world.
  1 KB (120 words) - 15:52, 9 August 2023
• An RPC DRAM Implementation for Energy-Efficient ASICs (1-2S)
  * Experience with HDLs (preferably SystemVerliog) such as taught in VLSI I * Must have visit VLSI2 in a previous semester or take it alongside the thesis
  8 KB (1,214 words) - 15:18, 9 July 2021
• Efficient TNN compression
  ...asks. Further, it has been shown that ternary feature maps are very smooth in the sense that moving from one pixel to a neighbouring pixel, very few bit Making use of this smoothness property in TNNs by implementing an efficient compression algorithm directly allows sav
  3 KB (438 words) - 08:41, 17 February 2021
• ISA extensions in the Snitch Processor for Signal Processing (M)
  ...ge quality, even on mobile devices, has lead to an increase in pixel count in smartphone cameras over the last decade [<nowiki/>[[#ref-Skafisk2017|1]]]. ...division instructions, or with custom user extensions. The processor used in MemPool is Snitch, a lightweight 32-bit RISC-V core developed at ETH Zurich
  9 KB (1,311 words) - 00:08, 13 March 2021
• Cell-Free mmWave Massive MIMO Communication
  ...on (6G) wireless systems are expected to exploit technologies already used in 5G such as massive MIMO and distributed antenna systems (DASs), but at high ...ving all users in the same time-frequency resource. Channels are estimated in each AP and the information is sent to the CPU, which performs detection, p
  7 KB (882 words) - 21:34, 13 July 2022
• Practical Reconfigurable Intelligent Surfaces (RIS)
  ...on, the expectation is to have more than 50 billions connected devices [1] in the network. To meet this demand, technologies that improve energy efficien ...able intelligent surfaces (RISs) are a promising solution that has emerged in recent years with the goal of altering electromagnetic fields by controllin
  8 KB (1,011 words) - 12:25, 16 November 2023
• Design-Space Exploration of Low-Resolution Matrix-Vector Multipliers
  ...le different ways to operate a matrix-vector product that we will consider in this project: The top row performs a naïve sum of outer products, while th ...ite their importance, there exist no systematic studies that have compared in-depth the different methods to compute a matrix-vector product.
  5 KB (653 words) - 11:08, 12 November 2020
• Low-Resolution 5G Beamforming Codebook Design
  ...control of the beam pattern to focus energy to/from a specific direction. In this example, we want to focus energy on the user located at 60°, while av ...an be tackled in several ways. For example, one can try to find a solution in real time (online), but then this solution needs to be found within a very
  6 KB (829 words) - 11:37, 12 November 2020
• Deep Unfolding of Iterative Optimization Algorithms
  ...xample of algorithm unrolled using Deep Unfolding. The parameters to learn in this example are delta and tau.]] ...gorithm to reach an adequate solution within a small number of iterations. In general, tuning these parameters is a time-consuming, manual effort that be
  6 KB (748 words) - 13:57, 12 November 2020
• ASIC Development of 5G-NR LDPC Decoder
  ...r-correcting codes, and therefore, have been adopted for data transmission in the 5G New-Radio (NR). VLSI implementation of the 5G-NR LDPC decoder is a c : An interest in digital signal processing
  3 KB (389 words) - 01:43, 10 February 2021
• Exploring schedules for incremental and annealing quantization algorithms
  Empirical evidence is supporting confidence in the fact that quantisation-aware training algorithms are necessary when tar Most algorithms in this family interpret the quantisation problem as an approximation task, wh
  6 KB (868 words) - 14:15, 15 February 2024
• Probabilistic training algorithms for quantized neural networks
  In recent theoretical research work, we have identified principled ways to com In contrast, the parameters of QNNs live in discrete spaces, where the meaning of gradient-based rules is less clear.
  4 KB (596 words) - 14:06, 15 February 2024
• FFT HDL Code Generator for Multi-Antenna mmWave Communication
  ...wever, the practical deployment of massive MIMO at mmWave faces challenges in terms of implementation costs and power consumption. ...ential to substantially reduce power consumption and implementation costs (in silicon area).
  3 KB (501 words) - 19:40, 16 November 2020
• Hypervisor Extension for Ariane (M)
  ...Machine-, Supervisor-, and User-Mode) and address translation as described in the RISC-V privileged ISA [2]. ...stems and cloud computing to allow running untrusted OSes or multiple OSes in parallel.
  3 KB (391 words) - 08:49, 21 June 2022
• Beamspace processing for 5G mmWave massive MIMO on GPU
  ...proach exploits channel sparsity during baseband processing, which results in reduced complexity compared to conventional antenna-domain processing. ...essing. Additionally, there is growing interest from GPU vendors to invest in the wireless communication industry. Hence, it is the perfect time to proto
  4 KB (620 words) - 00:16, 11 January 2022
• Autonomus Drones With Novel Sensors And Ultra Wide Band
  ...gies to enable high-precision and energy-efficient localization of sensors in the field as well as complete autonomous drones. The students will follow t * Indoor Localization using Ultra wide band and unique wake up radio designed in ITET will be used and evaluated by the students.
  5 KB (646 words) - 11:39, 30 November 2020
• Smart Patch For Heath Care And Rehabilitation
  ...in Zurich and Lausanne University Hospital and Neural Control of Movement in the Department of Health Sciences and Technology (D-HEST prof Nicole Wender ...re development IDE, PCB Design etc.) as well as to sensors that are unique in the world provided as early access by collaborative companies (such as STMi
  7 KB (1,033 words) - 16:24, 30 November 2020
• Matheus Cavalcante
  ...d my M.Sc. in integrated electronic systems from the Grenoble INP (Phelma) in 2018. I am currently pursuing a Ph.D. degree under the Digital Circuits and ==Projects In Progress==
  890 bytes (104 words) - 18:33, 8 December 2020
• Short Range Radars For Biomedical Application
  ...evices. However, their performance has not been tested and reported either in practical scenarios and very few works exploit machine learning to further ...ardware design, on the hardware-software co-design or in all of them (i.e. in the case of a master thesis) . Moreover, the project has the goal of acquir
  5 KB (692 words) - 12:46, 17 December 2021
• Bringup and Evaluation of an Energy-efficient Heterogeneous Manycore Compute Platform (1-2S)
  * RTL Design in SystemVerilog and Xilinx FPGA flow as taught in VLSI I * Programming in C
  11 KB (1,602 words) - 15:19, 9 July 2021
• Exploring feature selection and classification algorithms for ultra-low-power closed-loop systems for epilepsy control
  ...an abnormal signal that allows tailoring a therapy that is to be delivered in a closed-loop manner. ...s verified in real time. This technique is used in the control of epilepsy in closed-loop fashion, for example. Still, improving the efficiency of the de
  3 KB (388 words) - 09:25, 16 September 2021
• Efficient Synchronization of Manycore Systems (M/1S)
  ...C. The LR/SC instruction pair allows implementing locks and is widely used in lock-free algorithms. ..., one major drawback is their need for polling, which increases congestion in the interconnect, wastes energy, and can interfere with the thread holding
  12 KB (1,864 words) - 12:08, 29 August 2022
• Transforming MemPool into a CGRA (M)
  ...ge quality, even on mobile devices, has lead to an increase in pixel count in smartphone cameras over the last decade [[#ref-Skafisk2017|&#91;1&#93;]]. ...c array is the coarse grained reconfigurable architecture (CGRA). As shown in [[#fig:cgra|1]], these architectures implement a systolic array of configur
  13 KB (1,887 words) - 15:51, 17 November 2021
• Physical Implementation of Ara, PULP's Vector Machine (1-2S)
  In instruction-based programmable architectures, the key challenge is how to m In particular, multi-core designs, although highly flexible, do not explore th
  6 KB (916 words) - 15:25, 9 July 2021
• Towards the Ariane Desktop: Display Output for Ariane on FPGA under Linux (S/B/G)
  * Hand-In: June 28, 2021 ...feature set, making it possible to run RISC-V programs on custom hardware in real time.
  7 KB (1,122 words) - 15:21, 9 July 2021
• A Snitch-Based SoC on iCE40 FPGAs (1-2S/B)
  * Interest in computer architecture * Experience with HDLs as taught in ''VLSI I''
  8 KB (1,220 words) - 15:18, 9 July 2021
• Ottocore: A Minimal RISC-V Core Designed for Teaching (B/2G)
  * Interest in computer architecture * Preferably: Experience with HDLs as taught in VLSI I
  10 KB (1,521 words) - 15:21, 9 July 2021
• LLVM and DaCe for Snitch (1-2S)
  * Interest in code generation * Interest in or experience with the LLVM/Clang codebase
  11 KB (1,519 words) - 15:20, 9 July 2021
• Implementing A Low-Power Sensor Node Network
  In the current age of pervasive computing, the most common processing platform One of the key issues in enabling deployment of such Internet of Things devices is communication. Wh
  2 KB (366 words) - 15:53, 11 October 2021
• Efficient TNN Inference on PULP Systems
  * In the '''Hardware Track''', one student will implement TNN-specific ISA exten ...upport for the baseline architecture with the pre-existing ISA extensions. In a second step, the new TNN-specific ISA extensions developed on the hardwar
  5 KB (768 words) - 15:14, 4 August 2022
• Wearables in Fashion
  ...the spotlight of the Paris Fashion Week than in a dimly lit cellar office in the Gloriastrasse? Then we may have the right project for you! In this project, you will work in close cooperation with the University of Arts London to combine the latest
  3 KB (463 words) - 08:38, 23 November 2022
• Online Learning of User Features (1S)
  ...system. We want to explore how the user-specific features can be exploited in order to improve the accuracy of a small-footprint KWS model. ...ures belonging to a certain class. A schematic of a KWS system can be seen in Figure 1.
  11 KB (1,610 words) - 11:00, 14 November 2022
• Graph neural networks for epileptic seizure detection
  ..., 7, 8] using iEEG signals to detect two distinct states of brain activity in patients with epilepsy, i.e., interictal (= between seizures) and ictal (= ...with convolutional networks (GCNs), achieving state-of-the-art performance in public datasets [13].[14] proposed a temporal GCN to tackle the task of sei
  10 KB (1,306 words) - 19:58, 10 March 2024
• Securing Block Ciphers against SCA and SIFA
  ...y measure timing, power consumption or electromagnetic emanation of a chip in operation, and from these measurements try to extract secret information su ...p, e.g. by manipulating clock and reset inputs or with the use of a laser, in order to manipulate device operation.
  6 KB (849 words) - 18:43, 23 November 2021
• Implementation of an AES Hardware Processing Engine (B/S)
  ...an encryption algorithm used in many of today's applications. It operates in multiple rounds on fixed-size data blocks, converting a clear-text data str Depending on the progress, the component can furthermore be integrated into PULPissimo [4], and prepa
  2 KB (267 words) - 19:14, 6 December 2021
• Ibex: Tightly-Coupled Accelerators and ISA Extensions
  ...open source and it comes with an industry-grade verification environment. In addition, there are several configuration options such as optional support ...nsion interface has been started at ETH Zürich and University of Bologna. In the meantime, a working group has been formed connecting academia with indu
  6 KB (835 words) - 12:52, 27 April 2021
• Ultrasound
  ...working with [[:Category:Digital|digital]] ultrasound platforms developed in our lab, spanning from ultra-low-power wearable systems, to high-speed high ==Projects in Progress==
  797 bytes (90 words) - 16:37, 23 February 2022
• Novel Metastability Mitigation Technique
  ...tching clock. As a result, the circuit persists in an unstable equilibrium in metastability, which influences the correctness of the data and the power c ...against metastability, but it does not ensure the correctness of the data. In this project, the student will explore a novel approach that employs passiv
  2 KB (203 words) - 18:03, 16 December 2022
• Evaluating SoA Post-Training Quantization Algorithms
  ...optimal quantized weight values and/or activation quantization parameters. In this thesis, you will be implementing one or multiple of these algorithms a ...antization by Block Reconstruction ''(this is the current state of the art in post-training quantization)''
  6 KB (798 words) - 17:38, 1 September 2023
• Characterization techniques for silicon photonics-Lumiphase
  physical effect available in integrated photonics circuits has been a scientific and placed in the heart of our R&D activities and bridges topics related to materials
  4 KB (640 words) - 17:05, 16 September 2021
• Integrated silicon photonic structures-Lumiphase
  physical effect available in integrated photonics circuits has been a scientific and internship project is placed in the heart of our R&D activities and covers topics related
  4 KB (606 words) - 13:53, 23 June 2021
• Training and Deploying Next-Generation Quantized Neural Networks on Microcontrollers
  ...lass systems (MCUs) has seen intense attention from the research community in recent times, with the current state of the art being represented by MCUNet ...and deploy mixed-precision networks on PULP-based systems. More precisely, in this project, you will:
  5 KB (784 words) - 11:32, 29 October 2021
• Ternary Neural Networks for Face Recognition
  ...aset, optimize it for accuracy and prepare it for deployment on CUTIE, our in-house TNN accelerator. * [http://vis-www.cs.umass.edu/lfw/] Labelled Faces in the Wild
  3 KB (404 words) - 09:54, 8 March 2023
• Unconventional phase change memory device concepts for in-memory and neuromorphic computin
  ...ith the student. By the end of the thesis, the student will gain know-how in many aspects of the “chip” industry. ...ectronic phase-change synapses, broadly referred to by phase change memory in-memory computing.
  4 KB (546 words) - 13:07, 23 July 2021
• Fast Simulation of Manycore Systems (1S)
  In a quest for high-performance computing systems, few architectural models re ...s very tricky and an efficient model requires that the HDL code is written in an efficient manner. This thesis will profile MemPool’s Verilator model t
  10 KB (1,434 words) - 17:20, 2 August 2021
• Implementation of a Small and Energy-Efficient RISC-V-based Vector Accelerator (1M)
  == Status: In progress == ...age quality, even on mobile devices, has led to an increase in pixel count in smartphone cameras over the last decade.
  11 KB (1,609 words) - 10:00, 30 June 2022
• Universal Stream Semantic Registers for Snitch (1S)
  ...terns. Recent architectural extensions [1-2] propose handling such streams in hardware. This frees processors from explicitly computing addresses and iss ...(SSRs) [1]. These map memory streams directly to general-purpose registers in a RISC-V core, such that simply accessing a register loads or stores data.
  3 KB (425 words) - 17:32, 17 November 2021
• Efficient Memory Stream Handling in RISC-V-based Systems (M/1-2S)
  <!-- Efficient Memory Stream Handling in RISC-V-based Systems (M/1-2S) --> == Status: In Progress ==
  4 KB (557 words) - 16:14, 6 November 2022
• EEG artifact detection for epilepsy monitoring
  Epilepsy is a central nervous system disorder in which brain activity becomes abnormal, causing seizures or periods of unusu In this project, the student works with an EEG headband with dry and wet elect
  2 KB (313 words) - 15:25, 23 October 2023
• Novel Methods for Jammer Mitigation
  ...this project, we consider a scenario in which user equipments (UEs) try---in the presence of a jamming attack---to transmit data to a cellular basestati ...es in order to obtain the jammer-free UE signals. Things are not so simple in practice, however, when the jammer is "smart": Whenever the basestation tri
  6 KB (687 words) - 13:32, 10 May 2023
• EEG artifact detection with machine learning
  Epilepsy is a central nervous system disorder in which brain activity becomes abnormal, causing seizures or periods of unusu In this project, the student works with the Temple University (TUH) EEG Corpus
  4 KB (561 words) - 15:25, 23 October 2023
• Bluetooth Low Energy network with optimized data throughput
  ...i, F., & Chaudhry, F. J. (2010). Routing technique for inter slave bonding in Bluetooth scatternets formation. situations, 4, 6.)]] ...t sensors used to detect the movement during the scan, a Bluetooth network in master/slave (central/peripheral) configuration is needed.
  2 KB (249 words) - 17:18, 14 September 2021
• Every individual on the planet should have a real chance to obtain personalized medical therapy
  ...y forever. This project is not for the faint at heart. We operate in a highly competitive, globally distributed field. In particular, we are looking for one or several students who will contribute
  3 KB (387 words) - 17:04, 16 September 2021
• Probing the limits of fake-quantised neural networks
  However, in many applications where latency and data privacy are essential constraints, ...ing DNNs more efficient from several angles: optimising network topologies in terms of accuracy-per-parameter or accuracy-per-operation; reducing model s
  5 KB (659 words) - 14:08, 15 February 2024
• Feature Extraction for Speech Recognition (1S)
  ...be largely influenced by this choice, the aim of this project is to do an in-depth evaluation of feature extraction methods for KWS. ...xtraction, defining the former as the techniques applied to alter the data in order to emphasize or remove certain characteristics. To better understand
  12 KB (1,688 words) - 11:00, 14 November 2022
• Resource-Constrained Few-Shot Learning for Keyword Spotting (1S)
  ...eviate the limited access to data, yet its scope can feasibly be extended. In the current work, our goal is to obtain a few-shot learning pipeline target ...ent rates are low and data sets at each single institution are too small". In this occasions, one possible solution is few-shot learning.
  12 KB (1,869 words) - 17:37, 1 September 2023
• VLSI Design of an Asynchronous LDPC Decoder
  ...enerated layout of a simple processor implemented using ACT and fabricated in 65nm CMOS. ]] ...until now, not supported by EDA tools and, hence, enjoyed only limited use in practice.
  6 KB (725 words) - 17:36, 20 October 2021
• Mixed-Precision Neural Networks for Brain-Computer Interface Applications
  ...vices. Once fully functional, such a system would be of immeasurable value in the design of, e.g., motorized prostheses. ...rained edge devices such as microcontrollers, setting the state of the art in terms of efficiency for low-power motor imagery systems. However, the deplo
  8 KB (1,101 words) - 20:04, 10 March 2024
• Fault Tolerance
  ...ors, however they can induce a large overhead or are not always applicable in case data changes. Furthermore, for certain applications (e.g. handling imp ===Projects In Progress===
  2 KB (213 words) - 13:32, 6 October 2023
• Michael Rogenmoser
  ...h in 2020 and 2021, respectively. Since summer 2021, I am pursuing a Ph.D. in the Circuits and Systems group of Prof. Luca Benini. * Reliability in SoCs
  1 KB (150 words) - 17:30, 3 July 2023
• XNORLAX: Fused XNOR-LATCH Custom-Standard-Cell-Based Processing-in-Memory
  ...ions, such as machine learning, data mining, or 5G wireless communication, in which massive amounts of data need to be processed at high rates. ...der to break through the memory wall. The concept of PIM has been explored in many recent papers, but most of these works have focused on analog computat
  7 KB (933 words) - 19:29, 21 November 2021
• Robert Balas
  ===Projects In Progress=== category = In progress
  931 bytes (108 words) - 10:30, 22 November 2021
• Integration Of A Smart Vision System
  ...require power in the order of Watts, which precludes them from being used in a portable system. Furthermore, surveillance system typically do not come w In this project, a novel, distributed and energy-efficient surveillance system
  3 KB (433 words) - 15:36, 4 August 2022
• Flexfloat DL Training Framework
  ...round the minimal RISC-V Snitch integer core, only about 15 thousand gates in size, which is tightly coupled to accelerators such as an FPU or a DMA engi ===Status: In Progress ===
  4 KB (613 words) - 10:13, 2 November 2022
• Non-blocking Algorithms in Real-Time Operating Systems
  ...ep interrupt handling routines and critical sections as short as possible. In fact, the maximum interrupt latency (the worst-case) is largely determined ...oject is to look at data structures and task communication primitives used in FreeRTOS [4] [5], an open-source real-time operating system (RTOS) used by
  4 KB (508 words) - 18:59, 10 January 2022
• Real-Time Embedded Systems
  ...t also on the '''physical instant''' at which these results are produced. In fact, a real-time system changes its state as a function of physical time. ...ise it is a '''firm deadline'''. When missing a firm deadline results in a system failure (e.g. airplane sensor and autopilot systems, spacecrafts a
  4 KB (518 words) - 09:54, 10 January 2022
• Implementing DSP Instructions in Banshee (1S)
  <!-- Implementing DSP Instructions in Banshee (M/1S) --> * Experience or interest in learning Rust
  10 KB (1,428 words) - 13:31, 27 October 2022
• Accelerating Applications Relying on Matrix-Vector-Product-Like Operations
  [[File:IIP_ppac_row.png|800px|thumb|Array of processing-in-memory (PIM) bit-cells used by PPAC.]] ...ns. Depending on the application, such matrix-vector products are executed in different number formats (for example, GF2 arithmetic for forward-error cor
  7 KB (804 words) - 19:45, 21 November 2021
• HW/SW Safety and Security
  Please continue to [[Predictable_Execution]] if you are interested in timing predictable/safety critical embedded systems. Please continue to [[Real-Time Embedded Systems]] if you are interested in designing hardware or writing software for timing predictable embedded syst
  5 KB (614 words) - 09:49, 15 January 2024
• Low Resolution Neural Networks
  ...novel method developed in our group to reduce the complexity of inference in a network. ===Status: In Progress ===
  4 KB (520 words) - 14:52, 24 November 2021
• Wireless EEG Acquisition and Processing
  ...cessing unit that is not worn directly on the head. This must then be done in a way which the data is transmitted with short delay to the processing unit ...iable EEG signals during motion, transmitting them wirelessly with BioWolf in real time. An on-board accelerometer will empower the student to perform se
  3 KB (369 words) - 15:04, 20 July 2023
• Neural Architecture Search using Reinforcement Learning and Search Space Reduction
  ...nsumption, and latency. This results in the search space for manual design in being combinatorially large. A method of tackling the problem of manual des In this project, the student will explore NAS-subspaces generated using a meth
  5 KB (705 words) - 12:58, 16 January 2023
• New RVV 1.0 Vector Instructions for Ara
  [[Category:In progress]] ...1 processor. For example, “FUGAKU”, the most performant supercomputer in the world, is a vector processor!
  4 KB (580 words) - 11:37, 3 November 2023
• Passive Radar for UAV Detection using Machine Learning
  ...without actively transmitting any signals. The project will be carried out in collaboration with the Swiss startup company YOTASYS [2] and develop signal ===Status: In Progress ===
  5 KB (564 words) - 16:12, 9 February 2022
• Through Wall Radar Imaging using Machine Learning
  ...real-world data and software simulations. The project will be carried out in collaboration with the Swiss startup company YOTASYS [2]. ...and target differentiation have been identified as the main open questions in the field of TWRI. These challenges could potentially be addressed using ML
  5 KB (586 words) - 16:15, 9 February 2022
• Improved Collision Avoidance for Nano-drones
  In this project, we want to improve the reliability and accuracy of the obstac More in the details, the nano-drone platform is the Bitcraze Crazyflie 2.1 [4] exte
  4 KB (550 words) - 21:25, 15 February 2022
• Bridging QuantLab with LPDNN
  However, in many applications where latency and data privacy are essential constraints, ...ing DNNs more efficient from several angles: optimising network topologies in terms of accuracy-per-parameter or accuracy-per-operation; reducing model s
  7 KB (831 words) - 19:36, 12 January 2023
• Exploring NAS spaces with C-BRED
  ...ns, the deep learning research community has invested considerable efforts in neural architecture search (NAS). In a previous project, we developed Clustering-Based REDuction (C-BRED), a new
  6 KB (839 words) - 14:08, 15 February 2024
• Development of statistics and contention monitoring unit for PULP
  == Status: In progress == ...opular in the domain of '''Critical Real-Time Embedded Systems (CRTES)'''. In such systems, validation and verification (V&V) are a synergetic process th
  6 KB (869 words) - 14:47, 7 July 2023
• Development of an implantable Force sensor for orthopedic applications
  ...onitoring are a few examples that have brought transformational advantages in terms of accurate, individualized, and timely patient-centered care. Unfort The challenge in this project is to design an ultra low power bridge-to-digital converter fo
  3 KB (415 words) - 18:57, 13 February 2024
• A Novel Execution Scheme for Ultra-tiny CNNs Aboard Nano-UAVs
  ...ger in many practical applications such as aerial inspection or assistance in natural disaster areas, to name a few. ...will work on the State-of-the-Art open-source framework DORY [4], which is in charge of translating high-level PyTorch models into the final C implementa
  4 KB (489 words) - 14:33, 17 May 2022
• Low-Complexity MIMO Detection
  ===Status: In Progress === [[Category:In progress]]
  4 KB (503 words) - 13:54, 30 May 2022
• Forward error-correction ASIC using GRAND
  ===Status: In Progress === [[Category:In progress]]
  4 KB (470 words) - 18:16, 27 May 2022
• Low-power time synchronization for IoT applications
  ...chronization algorithm for multi-antenna wireless systems and implement it in hardware while minimizing area and power. ===Status: In Progress ===
  4 KB (492 words) - 10:55, 16 June 2022
• Quantum Transport Modeling of Interband Cascade Lasers (ICL)
  ...to find. To address this issue, quantum cascade lasers (QCL) were invented in 1994. The idea consists of taking advantage of intersubband electron-hole r ...interband cascade lasers based on the InAs-GaInSb material system with an in-house quantum transport solver called OMEN and to determine their performan
  3 KB (365 words) - 10:19, 31 May 2022
• Weak-strong massive MIMO communication with low-resolution ADCs
  ...the implementation of massive MIMO systems creates significant challenges in terms of power consumption, circuit complexity, and cost. To address these ...g for all users. Otherwise, stronger users would simply drown weaker users in quantization noise.
  5 KB (662 words) - 13:33, 10 May 2023
• Creating a Free and Open-Source Verification Environment for Our New DMA Engine (1-3S/B)
  [[Category:In progress]] == Status: In Progress ==
  2 KB (226 words) - 14:22, 27 February 2024
• Spatz grows wings: Physical Implementation of a Vector-Powered Manycore System (2S)
  ...age quality, even on mobile devices, has led to an increase in pixel count in smartphone cameras over the last decade. These image sensors, boasting tens In the quest to improve MemPool's peak performance, last semester, we proposed
  8 KB (1,239 words) - 12:36, 29 January 2024
• Analysis of Low-Power Wide Area Network Technologies for the Internet of Things
  ...1], mioty [2,3], or LTE-M [4]. The mioty standard is a promising candidate in terms of its excellent power consumption, security, range, and reliability. ===Status: In Progress ===
  5 KB (586 words) - 15:34, 11 July 2022
• Event-based navigation on autonomous nano-drones
  ...ing-based (DL) brains with novel event-based dynamic vision sensors (DVS). In this scenario, the DVS sensor will cover the role of a primary source of hi * Knowledge/experience in embedded programming and/or parallel programming is favorable.
  4 KB (505 words) - 18:25, 26 July 2022
• Bandwidth Efficient NEureka
  ...ompute budgets, task-specific acceleration has proven to be indispensable. In a heterogeneous subsystem with shared memory, it is very important to effic In this project, we will explore the bandwidth used by NEureka to execute a va
  4 KB (585 words) - 14:05, 15 February 2024
• Zephyr RTOS on PULP
  [[Category:In progress]] == Status: In progress ==
  3 KB (483 words) - 15:28, 19 February 2024
• Marco Bertuletti
  ...ted my B.Sc. and M.Sc. in Electronics Engineering at Politecnico di Milano in July 2019 and December 2021, respectively. Since March 2022, I am pursuing a Ph.D. in the Circuits and Systems group of Prof. Luca Benini.
  2 KB (204 words) - 11:29, 23 December 2023
• A Flexible FPGA-Based Peripheral Platform Extending Linux-Capable Systems on Chip (1-3S/B)
  ...'Integration of PAPER:''' Integrate PAPER, a DVI/HDMI peripheral developed in previous student projects. * '''U-Boot and Linux Device Drivers:''' Integrate the necessary drivers in U-Boot and Linux to use the integrated peripherals
  3 KB (416 words) - 10:49, 25 January 2024
• On-Device Training Sparse Sub-Tensor Update Scheme Optimization for CNN-based tasks (SA or MA)
  ...on inference, under the train-then-deploy paradigm. However, this results in a device unable to face real-life phenomena such as data distribution shift ...ce still requires a considerable amount of memory, a significant challenge in the context of tightly memory constrained devices such as Microcontrollers
  7 KB (935 words) - 13:29, 2 November 2023
• Alias-Free Oscillator Synchronization for Arbitrary Waveforms
  ...g-free synchronization of a simple sine waveform is still ongoing research in 2022 [3]. ...cillator (BFO) project, to enable aliasing-free oscillator synchronization in real-time.
  5 KB (577 words) - 09:48, 5 October 2022
• Virtual Memory Ara
  [[Category:In progress]] ...1 processor. For example, “FUGAKU”, the most performant supercomputer in the world, is a vector processor!
  5 KB (769 words) - 11:38, 3 November 2023
• Audio Visual Speech Separation and Recognition (1S/1M)
  ...ang2018|&#91;7&#93;]]. This aim to mimic the cocktail party effect present in humans, which denotes our capability of focusing on an individual conversat In the current work, we investigate the potential of performing Audio-Visual S
  11 KB (1,467 words) - 14:15, 1 March 2024
• Runtime partitioning of L1 memory in Mempool (M)
  <!-- Runtime partitioning of L1 memory in Mempool (M) --> [[Category:In progress]]
  3 KB (490 words) - 10:38, 2 November 2023
• GDBTrace: A Post-Simulation Trace-Based RISC-V GDB Debugging Server (1S)
  [[Category:In progress]] == Status: In progress ==
  8 KB (1,186 words) - 11:49, 13 March 2024
• Radiation Testing of a PULP ASIC
  [[Category:In progress]] ...ned to allow us to sample as much data as possible during the short window in which we will have access to the facilities.
  3 KB (454 words) - 14:59, 25 October 2023
• Towards a Technology-independent and Synthesizable AXI4 Performance Monitoring and Throttling Unit (1-2S)
  == Status: In Progress == [[Category:In Progress]]
  2 KB (335 words) - 13:58, 27 October 2022
• Big Data Analytics Benchmarks for Ara
  [[Category:In progress]] == Status: In-Progress ==
  3 KB (470 words) - 11:34, 3 November 2023
• Creating A Technology-independent USB1.0 Host Implementation Targetting ASICSs (1-3S/B)
  [[Category:In progress]] ...requires a lot of effort. A simple solution to this problem was introduced in 1996 with USB 1.0.
  2 KB (220 words) - 09:27, 3 November 2023
• Molecular Binding Kinetics Modelling of NO2 on Graphene/hBN Heterostructure
  ...face exposed to molecules, as shown above, allows to explore its potential in gas sensor applications. To study the influence of certain reactive molecu • Strong interest in semiconductor device physics and simulation
  2 KB (302 words) - 18:47, 10 November 2022
• Vector-based Parallel Programming Optimization of Communication Algorithm (1-2S/B)
  ...1024 Snitch cores and 4096 banks of shared tightly coupled L1 data-memory. In this project, we will exhibit considerable DLP and implement the typical ke In the initial phase of the project, your primary objective will be to acquain
  6 KB (775 words) - 11:57, 31 October 2023
• Design of an Area-Optimized Soft-Error Resilient Processing Core for Safety-Critical Systems (1M)
  ...soft error resilient processing core, leveraging the concept of repetition in-time. ...ted, cycle-by-cycle, by a majority voter. If the voter detects discrepancy in the results, it alerts the system that a system failure happened and recove
  5 KB (752 words) - 13:23, 24 October 2023
• Design of a Reconfigurable Vector Processor Cluster for Area Efficient Radar Processing (1M)
  ...units and scalar processors into a heterogeneous fabric is very expensive in terms of area and power budget, especially when we consider embedded system ...efficient, radar signal processing. The vector cluster will operate either in scalar or vector mode and will efficiently sustain control tasks as well as
  5 KB (651 words) - 20:42, 22 November 2022
• Design of an Energy-Efficient Ethernet Interface for Linux-capable Systems
  ...s (WAN) [1]. It was commercially introduced in 1980 and first standardized in 1983 as IEEE 802.3 ...roject by investigating the challenges and opportunities of using Ethernet in real-time systems for automotive-oriented applications [3].
  5 KB (631 words) - 09:28, 3 November 2023
• Design of a CAN Interface to Enable Reliable Sensors-to-Processors Communication for Automotive-oriented Embedded Applications (1M)
  ...utomation and other areas of networked embedded control, with applications in diverse products such as combustion and electric vehicles, agriculture, avi ...vices, including by the transmitting device. More information can be found in [1].
  4 KB (554 words) - 09:28, 3 November 2023
• Predict eye movement through brain activity
  ...approaches to this problem is a measurement of brain activity with sensors in glasses or a headband. The advertised project aims to provide a proof of co [[File:Headband v2 in-min.jpeg|thumb|right|200px]]
  3 KB (420 words) - 15:27, 23 October 2023
• Super Resolution Radar/Imaging at mm-Wave frequencies
  ...l-time precise localization between transmitters (TXs) and receivers (RXs) in dynamic mobile applications. ...alignment and reliable communication, especially in mobile wireless links. In parallel, array-based mm-Wave radars require large arrays for precise angul
  2 KB (326 words) - 11:44, 14 February 2023
• Machine Learning Assisted Direct Synthesis of Passive Networks
  ...of the passive network structure grows, simulation times may easily reach in the excess of several hours. This prevents the designer from performing opt ...or automating the design of transformers to achieve optimal power matching in electrical systems. The solution will be based on a mathematical model that
  4 KB (524 words) - 11:44, 14 February 2023
• Continual Learning for Adaptive EEG Monitoring in Epileptic Seizure Detection
  <!-- Continual Learning for Adaptive EEG Monitoring in Epileptic Seizure Detection: A Wearable Approach --> In this project, the student will focus on using Continual Learning to improve
  8 KB (1,271 words) - 15:04, 20 July 2023
• Audio Visual Speech Separation (1S/1M)
  ...ang2018|&#91;7&#93;]]. This aim to mimic the cocktail party effect present in humans, which denotes our capability of focusing on an individual conversat In the current work, we investigate the potential of performing Audio-Visual S
  9 KB (1,283 words) - 17:44, 1 September 2023
• Charge and heat transport through graphene nanoribbon based devices
  ...However, their use as quantum heat engines has not been demonstrated yet. Progress on the experimental, as well as on the theoretical side is required to gain Interest in quantum transport theory and nano-device modeling.
  3 KB (460 words) - 17:28, 20 February 2023
• Efficient Banded Matrix Multiplication for Quantum Transport Simulations
  ...sparse matrices are unstructured. Therefore, using these libraries results in sub-optimal performance and/or much higher memory usage than necessary comp ===Status: In Progress ===
  3 KB (446 words) - 15:56, 1 May 2023
• Artificial Reverberation for Embedded Systems
  ...ounts of memory to model the propagation of acoustic waves in large rooms. In fact, storing 24-bit stereo samples at a 96kHz sampling rate for one second ...fabricated chip should be able to apply the "tiny" reverberation algorithm in real-time to signals sampled at a stereo ADC.
  5 KB (578 words) - 12:39, 14 June 2023
• On-Board Software for PULP on a Satellite
  [[Category:In progress]] ...enos aims to be onboard, but still requires software both to run its tests in space and interface with the main On-Board Computer.
  3 KB (383 words) - 14:29, 18 February 2024
• A FPGA-based data streaming system that enables real-time monitoring of cell culture and neuroactivities
  ...information such as the drug efficacy and toxicity can be readily assessed in vitro. As such, the development of CMOS biosensors for cell monitoring and : Programming experience in Python and C++
  6 KB (720 words) - 16:27, 27 September 2023
• Real-time Linux on RISC-V
  [[Category:In Progress]] == Status: In Progress ==
  3 KB (460 words) - 15:29, 19 February 2024
• Wireless In Action Data Streaming in Ski Jumping (1 B/S)
  ...ta with a body-worn sensor node and transmits the information to the coach in real time. This requires reliable wireless communication with sufficient ba ===Status: In Progress ===
  6 KB (688 words) - 12:15, 23 July 2023
• Cycle-Accurate Event-Based Simulation of Snitch Core
  ...hitectures. In this scenario, high-level simulators play an essential role in breaking the speed and design effort bottlenecks of cycle-accurate simulato ...with the flexibility of Python configuration scripts. An overview is shown in the Figure. GVSoC combines the following concepts:
  14 KB (2,018 words) - 22:54, 23 November 2023
• Smart Goggles for Visual In-Action Feedback in Ski Jumping (1 M 1-2B/S)
  In ski jumping, low repetition rates of jumps limit effectiveness of training. ===Status: In Progress ===
  6 KB (735 words) - 12:15, 23 July 2023
• On - Device Continual Learning for Seizure Detection on GAP9
  == Status: In progress == [[Category:In progress]]
  9 KB (1,326 words) - 11:16, 12 October 2023
• Pretraining Foundational Models for EEG Signal Analysis Using Open Source Large Scale Datasets
  ...r capturing the brain's electrical activity, playing an indispensable role in both neurological research and clinical diagnostics. It acts as a portal to ...s inherent in these signals. This initiative recognizes a significant void in AI applications: while foundation models have revolutionized areas like ima
  7 KB (891 words) - 17:04, 18 December 2023
• Skin coupling media characterization for fitnesstracker applications (1 B/S)
  ...uman anatomy, which is hard to achieve with current transducer technology. In this work, we develop strategies and prototypes to advance ultrasound into ===Status: In Progress ===
  5 KB (631 words) - 12:43, 23 July 2023
• Integrating Ultrasound Technology into a Fitness Tracking Device (1M, 2 B/S)
  ...uman anatomy, which is hard to achieve with current transducer technology. In this work, we develop strategies and prototypes to advance ultrasound into * Showing participation in non-curricular analog/digital projects is a plus.
  7 KB (903 words) - 10:04, 24 July 2023
• Skin Coupling Media Characterization For Fitness Tracker Applications (1 B/S)
  ...uman anatomy, which is hard to achieve with current transducer technology. In this work, we develop strategies and prototypes to advance ultrasound into ===Status: In Progress ===
  5 KB (631 words) - 10:07, 24 July 2023
• Implementation of a Cache Reliability Mechanism (1S/M)
  [[Category:In progress]] == Status: In Progress ==
  2 KB (267 words) - 14:22, 27 February 2024
• Transformer Deployment on Heterogeneous Many-Core Systems
  ...processing, and machine learning are offloaded to accelerators, resulting in improved performance and energy efficiency. .... The self-attention mechanism is a key component of the model but results in a high computational load and a complicated operational scheme. For this re
  6 KB (858 words) - 14:52, 23 October 2023
• Fault-Tolerant Floating-Point Units (M)
  [[Category:In progress]] == Status: In Progress ==
  2 KB (273 words) - 14:23, 27 February 2024
• Audio Visual Speech Recognition (1S/1M)
  ...ang2018|&#91;7&#93;]]. This aim to mimic the cocktail party effect present in humans, which denotes our capability of focusing on an individual conversat In the current work, we investigate the potential of performing Audio-Visual S
  9 KB (1,283 words) - 18:08, 1 September 2023
• Extreme-Edge Experience Replay for Keyword Spotting
  ...s)[[#ref-Hemati2023|&#91;2&#93;]] are added directly in the target device. In such contexts, on-device continual learning -- both domain incremental and ...ce catastrophic forgetting by maintaining a subset of already-seen samples in a memory buffer, also called reservoir. During the adaptation stage, the su
  9 KB (1,226 words) - 13:56, 15 February 2024
• Learning at the Edge with Hardware-Aware Algorithms
  ...le to face real-life phenomena such as data distribution shifts or changes in the target classes. ...trated on MCUs with straightforward memory hierarchies like ARM Cortex-M4. In contrast, cutting-edge Edge processors such as GAP9 [[#ref-gwt|&#91;2&#93;]
  7 KB (942 words) - 13:59, 15 February 2024
• Reliability by Switching the Embedded ISA in Ibex (1-2S/B/1M)
  <!-- Reliability by Switching the Embedded ISA in Ibex --> ...PULP group, we have started developing reliable hardware designed for use in space, where high levels of radiation have a significant impact on the corr
  2 KB (259 words) - 11:55, 18 December 2023
• Designing a Scalable Miniature I/O DMA (1-2B/1-3S/M)
  [[Category:In progress]] == Status: In Progress ==
  1 KB (192 words) - 14:24, 27 February 2024
• Physical Implementation of ITA (2S)
  [[Category:In progress]] Transformers [1], initially popularized in Natural Language Processing (NLP), have found applications beyond and are n
  4 KB (577 words) - 10:52, 12 December 2023
• 3D Matrix Multiplication Unit for ITA (1S)
  [[Category:In progress]] Transformers [1], initially popularized in Natural Language Processing (NLP), have found applications beyond and are n
  3 KB (485 words) - 10:52, 12 December 2023
• Streaming Layer Normalization in ITA (M/1-2S)
  <!-- Streaming Layer Normalization in ITA (M/1-2S) --> [[Category:In progress]]
  4 KB (511 words) - 12:38, 21 December 2023
• Vector-based Manycore HPC Cluster Exploration for 5G Communication Algorithm (1-2M)
  [[Category:In progress]] == Status: In Progress ==
  6 KB (844 words) - 11:41, 31 October 2023
• A RISC-V fault-tolerant many-core accelerator for 5G Non-Terrestrial Networks (1-2S/B)
  [[Category:In progress]] == Status: In progress ==
  3 KB (342 words) - 13:02, 12 February 2024
• Design of a Scalable High-Performance and Low-Power Interface Based on the I3C Protocol (1-3S/B)
  ...ion interfaces are required to control and transmit data with high speeds, in energy-saving and space-saving designs. ...t-benches, (following the UVM methodology [4] if the student is interested in this topic). Eventually, the project can be completed with the integration
  5 KB (775 words) - 17:17, 18 December 2023
• Creating an At-memory Low-overhead Bufferless Matrix Transposition Accelerator (1-3S/B)
  ...ible family of DMA engines, called iDMA [1]. iDMA is the cluster-level DMA in both the Snitch and PULP clusters. When implemented as this cluster-level e ...a matrix, an accelerator requires a huge internal buffer to read the data in a dense format, reshuffle it, and write it out again as a dense stream. Thi
  2 KB (314 words) - 10:27, 3 November 2023
• Efficient collective communications in FlooNoC (1M)
  <!-- Efficient collective communications in FlooNoC (1M) --> [[Category:In progress]]
  6 KB (897 words) - 19:52, 22 February 2024
• Investigation of Quantization Strategies for Retentive Networks (1S)
  [[Category:In progress]] == Status: In Progress ==
  6 KB (844 words) - 14:28, 18 February 2024
• Advanced Data Movers for Modern Neural Networks
  In the current era of AI everything, efficient neural network inference has be In this uniquely vertical project, the student will have the opportunity to
  3 KB (418 words) - 16:18, 23 November 2023
• ASIC implementation of a beamspace massive MIMO-OFDM detector for 5G/6G
  ...ss channels, orthogonal frequency division multiplexing (OFDM) is employed in the 5G standard [1], like many other standards such as LTE and IEEE 802.11 * Large number of basestation (BS) antennas in massive MIMO results in large data dimensions to be processed, which makes the detection task compl
  7 KB (939 words) - 15:51, 28 April 2024
• Deep Learning Based Anomaly Detection in ECG Signals Using Foundation Models
  ...l representation of the heart's electrical activity. ECGs are instrumental in identifying various heart conditions, including arrhythmias, myocardial inf ...for detecting anomalies in ECG signals, utilizing the latest advancements in deep learning and foundational models. This system is designed to leverage
  6 KB (803 words) - 19:08, 14 December 2023
• Exploratory Development of a Unified Foundational Model for Multi Biosignal Analysis
  ...le in diagnosing, monitoring, and understanding various health conditions. In the realm of biosignal interpretation, diverse types of signals like electr ...a more comprehensive and holistic view of a patient's physiological state. In medical research, this approach can facilitate a more integrated understand
  6 KB (811 words) - 18:00, 15 December 2023
• A RISC-V ISA Extension for Scalar Chaining in Snitch (M)
  <!-- A RISC-V ISA Extension for Scalar Chaining in Snitch (1M) --> [[Category:In progress]]
  7 KB (1,153 words) - 18:58, 21 April 2024
• ASR-Waveformer
  == Status: In Progress == [[Category:In progress]]
  10 KB (1,368 words) - 11:50, 18 March 2024
• On-Device Federated Continual Learning on Nano-Drone Swarms
  This approach is crucial for applications in dynamic environments where data distributions change or new tasks emerge. This is crucial for privacy-sensitive applications and in scenarios where data cannot be centralized due to bandwidth or regulatory c
  10 KB (1,377 words) - 00:49, 29 February 2024
• On-Device Learnable Embeddings for Acoustic Environments
  ...oom. We want to explore how environment-specific features can be exploited in order to improve the accuracy of a small-footprint KWS model. ...aring their interests against clusters formed by other users’ interests. In other words, the engine adapts itself based on user-specific features.
  11 KB (1,500 words) - 01:10, 29 February 2024

View (previous 250 | next 250) (20 | 50 | 100 | 250 | 500)