Martin Kumm
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Martin Kumm has a total of 122 co-authors. Here are the top co-authors by number of shared publications:
Peter Zipf
Digital Technology Group, University of Kassel, Kassel, Germany
4 shared publications- Peter Zipf
University of Kassel, Germany
4 shared publications - Anastasia Volkova
Sorbonne Universités, UPMC Univ Paris 06, UMR 7606, LIP6, F-75005, Paris, France
2 shared publications - Konrad Moller
Digital Technology Group University of Kassel, Germany
2 shared publications - Mario Garrido
Department of Electronic Engineering, ETSI de Telecomunicación, Universidad Politécnica de Madrid, Madrid, Spain
2 shared publications Martin Hardieck
Digital Technology Group, University of Kassel, 34121 Kassel, Germany
2 shared publications
Latest Publications
- Open Access
- 29 October 2026
- Published by Wiley in Geoscience Data Journal
- Vol. 13 (4)
The Wessex Convection Experiment ( WesCon )–Observing the Evolving Structures of Turbulence ( WOEST ): Coordinated Aircraft and Ground‐Based Observations of Turbulence and Convection in Southwest England During Summer 2023
The goal of the ParaChute programme (Parametrising Convective turbulence at Hectometric scales and Understanding the Turbulent Environment) is to improve the representation of turbulence in km‐scale and 100‐m‐scale weather and climate models. The field component of ParaChute was composed of two overlapping and coordinated campaigns during summer 2023 in southern England: the Met Office–funded Wessex Convection Experiment (WesCon) and the NERC‐funded WesCon Observing the Evolving Structures of Turbulence (WOEST). WesCon–WOEST deployed an array of continuous ground‐based observing systems, augmented by additional instrumentation during 30 Intensive Observing Periods (IOPs). The instrumentation collected surface‐based in situ measurements, remotely sensed ground‐based measurements (e.g., radars, lidars, vertically‐profiling instrumentation), and aerial observations from two crewed research aircraft and one uncrewed aerial system. This article synthesises the observations these two field campaigns collected, describes the instrumentation, describes the resulting datasets and their availability, and showcases an illustrative case study for a day where the development of a convective boundary layer led to convective storms. The datasets from this study, together with operational and experimental numerical weather prediction model outputs, will be used to develop new understanding of the turbulent processes on days that produce convection, supporting complementary goals in designing new turbulence parametrisations.
- 29 October 2026
- Published by Wiley in Geoscience Data Journal
- Vol. 13 (4)
Open AccessThe Wessex Convection Experiment ( WesCon )–Observing the Evolving Structures of Turbulence ( WOEST ): Coordinated Aircraft and Ground‐Based Observations of Turbulence and Convection in Southwest England During Summer 2023
The goal of the ParaChute programme (Parametrising Convective turbulence at Hectometric scales and Understanding the Turbulent Environment) is to improve the representation of turbulence in km‐scale and 100‐m‐scale weather and climate models. The field component of ParaChute was composed of two overlapping and coordinated campaigns during summer 2023 in southern England: the Met Office–funded Wessex Convection Experiment (WesCon) and the NERC‐funded WesCon Observing the Evolving Structures of Turbulence (WOEST). WesCon–WOEST deployed an array of continuous ground‐based observing systems, augmented by additional instrumentation during 30 Intensive Observing Periods (IOPs). The instrumentation collected surface‐based in situ measurements, remotely sensed ground‐based measurements (e.g., radars, lidars, vertically‐profiling instrumentation), and aerial observations from two crewed research aircraft and one uncrewed aerial system. This article synthesises the observations these two field campaigns collected, describes the instrumentation, describes the resulting datasets and their availability, and showcases an illustrative case study for a day where the development of a convective boundary layer led to convective storms. The datasets from this study, together with operational and experimental numerical weather prediction model outputs, will be used to develop new understanding of the turbulent processes on days that produce convection, supporting complementary goals in designing new turbulence parametrisations.
- Open Access
- 19 August 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Reconfigurable Technology and Systems
Evaluating Monotonic Functions using Binary Search
This work addresses the accurate (correctly rounded) evaluation of arbitrary functions. The simplest solution is to use a Look-Up Table (LUT), with a storage requirement (hence area) that grows linearly with the output word size, but exponentially with the input word size. This limits its applicability to small input word sizes. For monotonic functions, a binary search in the output domain enables implementations whose area scales linearly with the input word size and exponentially with the output word size. This article introduces several novel architectures exploiting this idea, investigates their efficient mapping to FPGA micro-architecture using features such as \(k\) -input LUT and fast carry logic, compares them to the LUT approach in area and delay, and demonstrates an application to a Gaussian pseudo-random number generator.- 19 August 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Reconfigurable Technology and Systems
Open AccessEvaluating Monotonic Functions using Binary Search
This work addresses the accurate (correctly rounded) evaluation of arbitrary functions. The simplest solution is to use a Look-Up Table (LUT), with a storage requirement (hence area) that grows linearly with the output word size, but exponentially with the input word size. This limits its applicability to small input word sizes. For monotonic functions, a binary search in the output domain enables implementations whose area scales linearly with the input word size and exponentially with the output word size. This article introduces several novel architectures exploiting this idea, investigates their efficient mapping to FPGA micro-architecture using features such as \(k\) -input LUT and fast carry logic, compares them to the LUT approach in area and delay, and demonstrates an application to a Gaussian pseudo-random number generator. - 5 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Artificial Intelligence
- Vol. PP (4) , 1-14
Continuous-Flow Data-Rate-Aware CNN Inference on FPGA
Among hardware accelerators for deep-learning inference, data flow implementations offer low latency and high throughput capabilities. In these architectures, each neuron is mapped to a dedicated hardware unit, making them wellsuited for field-programmable gate array (FPGA) implementation. Previous unrolled implementations mostly focus on fully connected networks because of their simplicity, although it is well known that convolutional neural networks (CNNs) require fewer computations for the same accuracy. When observing the data flow in CNNs, pooling layers and convolutional layers with a stride larger than one, the number of data at their output is reduced with respect to their input. This data reduction strongly affects the data rate in a fully parallel implementation, making hardware units heavily underutilized unless it is handled properly. This work addresses this issue by analyzing the data flow of CNNs and presents a novel approach to designing data-rate-aware, continuous-flow CNN architectures. The proposed approach ensures a high hardware utilization close to 100% by interleaving low data rate signals and sharing hardware units, as well as using the right parallelization to achieve the throughput of a fully parallel implementation. The results show that a significant amount of the arithmetic logic can be saved, which allows implementing complex CNNs like MobileNet on a single FPGA with high throughput.
- 5 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Artificial Intelligence
- Vol. PP (4) , 1-14
Continuous-Flow Data-Rate-Aware CNN Inference on FPGA
Among hardware accelerators for deep-learning inference, data flow implementations offer low latency and high throughput capabilities. In these architectures, each neuron is mapped to a dedicated hardware unit, making them wellsuited for field-programmable gate array (FPGA) implementation. Previous unrolled implementations mostly focus on fully connected networks because of their simplicity, although it is well known that convolutional neural networks (CNNs) require fewer computations for the same accuracy. When observing the data flow in CNNs, pooling layers and convolutional layers with a stride larger than one, the number of data at their output is reduced with respect to their input. This data reduction strongly affects the data rate in a fully parallel implementation, making hardware units heavily underutilized unless it is handled properly. This work addresses this issue by analyzing the data flow of CNNs and presents a novel approach to designing data-rate-aware, continuous-flow CNN architectures. The proposed approach ensures a high hardware utilization close to 100% by interleaving low data rate signals and sharing hardware units, as well as using the right parallelization to achieve the throughput of a fully parallel implementation. The results show that a significant amount of the arithmetic logic can be saved, which allows implementing complex CNNs like MobileNet on a single FPGA with high throughput.
- 16 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 71 (1) , 249-261
Bit-Level Optimized Constant Multiplication Using Boolean Satisfiability
Multiplierless constant multiplication using bit-shifts, additions and subtractions has been an active research topic in the last decades. The multiplication with multiple constants, known as the multiple constant multiplication (MCM) problem, is of special interest because of its practical relevance, notably for digital filter implementation. In this work we propose to use the speed of modern Boolean satisfiability (SAT) solvers to find fast and optimal solutions. The solutions are optimal either with respect to the adder count or the bit level cost. In contrast to previous approaches, we also consider negative fundamentals that are sometimes cheaper to realize than their positive counterparts leading to more compact hardware implementations. Our experiments show that our approach is able to find optimal single constant multiplication (SCM) and MCM circuits for practically relevant test instances in reasonable time. We also prove the necessity for the post-add right shift operation for SCM. Using our SAT formulation to enumerate all possible implementations for some of our test instances we show the importance of considering bit-level costs and negative fundamentals when solving MCM problems.
- 16 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 71 (1) , 249-261
Bit-Level Optimized Constant Multiplication Using Boolean Satisfiability
Multiplierless constant multiplication using bit-shifts, additions and subtractions has been an active research topic in the last decades. The multiplication with multiple constants, known as the multiple constant multiplication (MCM) problem, is of special interest because of its practical relevance, notably for digital filter implementation. In this work we propose to use the speed of modern Boolean satisfiability (SAT) solvers to find fast and optimal solutions. The solutions are optimal either with respect to the adder count or the bit level cost. In contrast to previous approaches, we also consider negative fundamentals that are sometimes cheaper to realize than their positive counterparts leading to more compact hardware implementations. Our experiments show that our approach is able to find optimal single constant multiplication (SCM) and MCM circuits for practically relevant test instances in reasonable time. We also prove the necessity for the post-add right shift operation for SCM. Using our SAT formulation to enumerate all possible implementations for some of our test instances we show the importance of considering bit-level costs and negative fundamentals when solving MCM problems.
- 19 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 72 (5) , 1261-1273
Towards Globally Optimal Design of Multipliers for FPGAs
The design of a multiplier typically consists of three steps: (1) partial product generation, (2) compressor tree design and (3) the selection of the final adder. Conventionally, these three steps are performed consecutively. However, when targeting FPGAs, there are many possibilities in all three design steps that heavily influence each other. This proposal presents for the first time a holistic optimization, combining all three optimization steps yielding a minimum amount of look-up-tables (LUTs) while it can also guarantee the minimal number of (pipeline) stages. An ILP-formulation for the determination of a combined, globally optimal solution for the multiplier tiling, compressor tree generation and final adder selection is proposed. With globally optimal we mean that the best solution is found for a given set of sub-multipliers for partial product generation, compressors and final adder.This allows to improve the quality and evaluate the limitations of existing heuristic 3-step approaches. It is shown experimentally for the example of Xilinx FPGAs, that globally optimal solutions can be obtained for multiplier sizes of practical relevance, leading to significant LUT reductions. Additional packing density experiments show that a significantly larger number of multiplier instances can be mapped to the same device.
- 19 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 72 (5) , 1261-1273
Towards Globally Optimal Design of Multipliers for FPGAs
The design of a multiplier typically consists of three steps: (1) partial product generation, (2) compressor tree design and (3) the selection of the final adder. Conventionally, these three steps are performed consecutively. However, when targeting FPGAs, there are many possibilities in all three design steps that heavily influence each other. This proposal presents for the first time a holistic optimization, combining all three optimization steps yielding a minimum amount of look-up-tables (LUTs) while it can also guarantee the minimal number of (pipeline) stages. An ILP-formulation for the determination of a combined, globally optimal solution for the multiplier tiling, compressor tree generation and final adder selection is proposed. With globally optimal we mean that the best solution is found for a given set of sub-multipliers for partial product generation, compressors and final adder.This allows to improve the quality and evaluate the limitations of existing heuristic 3-step approaches. It is shown experimentally for the example of Xilinx FPGAs, that globally optimal solutions can be obtained for multiplier sizes of practical relevance, leading to significant LUT reductions. Additional packing density experiments show that a significantly larger number of multiplier instances can be mapped to the same device.
- Open Access
- 1 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 475-478
Author Index
- 1 August 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 475-478
Open AccessAuthor Index
- 30 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 42 (2) , 658-671
Design of Optimal Multiplierless FIR Filters With Minimal Number of Adders
This work presents two novel methods that simultaneously optimize both the design of a finite impulse response (FIR) filter and its multiplierless hardware implementation. We use integer linear programming (ILP) to minimize the number of adders used to implement a direct/transposed FIR filter adhering to a given frequency specification. The proposed algorithms work by either fixing the number of adders used to implement the products (multiplier block adders) or by bounding the adder depth (AD) used for these products. The latter can be used to design filters with minimal AD for low-power applications. In contrast to previous multiplierless FIR filter approaches, the methods introduced here ensure adder count optimality. We perform extensive numerical experiments, which demonstrate that our simultaneous filter design approach yields results that are in many cases on par or better than those in the literature.
- 30 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 42 (2) , 658-671
Design of Optimal Multiplierless FIR Filters With Minimal Number of Adders
This work presents two novel methods that simultaneously optimize both the design of a finite impulse response (FIR) filter and its multiplierless hardware implementation. We use integer linear programming (ILP) to minimize the number of adders used to implement a direct/transposed FIR filter adhering to a given frequency specification. The proposed algorithms work by either fixing the number of adders used to implement the products (multiplier block adders) or by bounding the adder depth (AD) used for these products. The latter can be used to design filters with minimal AD for low-power applications. In contrast to previous multiplierless FIR filter approaches, the methods introduced here ensure adder count optimality. We perform extensive numerical experiments, which demonstrate that our simultaneous filter design approach yields results that are in many cases on par or better than those in the literature.
- 22 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 70 (1053587X) , 1673-1686
Hardware-Aware Design of Multiplierless Second-Order IIR Filters With Minimum Adders
In this work we optimally solve the problem of multiplierless design of second-order Infinite Impulse Response filters with minimum number of adders. Given a frequency specification, we design a stable direct form filter with hardware-aware fixed-point coefficients where all multiplications are replaced by bit shifts and additions. The coefficient design, quantization and implementation, typically conducted independently, are now gathered into one global optimization problem, modeled through integer linear programming and efficiently solved using generic solvers. The optimal filters are implemented within the FloPoCo IP core generator and synthesized for field programmable gate arrays (FPGAs) and application specific integrated circuits (ASICs). With respect to state-of-the-art three-step filter design methods, our one-step design approach achieves, on average, 48% reduction in number of lookup tables, 27% delay reduction and 57% reduction in power on FPGAs. ASICs experiment illustrate similar 48% reduction in circuit area, 27% delay reduction and 65% power reduction for a 14nm ASIC.
- 22 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 70 (1053587X) , 1673-1686
Hardware-Aware Design of Multiplierless Second-Order IIR Filters With Minimum Adders
In this work we optimally solve the problem of multiplierless design of second-order Infinite Impulse Response filters with minimum number of adders. Given a frequency specification, we design a stable direct form filter with hardware-aware fixed-point coefficients where all multiplications are replaced by bit shifts and additions. The coefficient design, quantization and implementation, typically conducted independently, are now gathered into one global optimization problem, modeled through integer linear programming and efficiently solved using generic solvers. The optimal filters are implemented within the FloPoCo IP core generator and synthesized for field programmable gate arrays (FPGAs) and application specific integrated circuits (ASICs). With respect to state-of-the-art three-step filter design methods, our one-step design approach achieves, on average, 48% reduction in number of lookup tables, 27% delay reduction and 57% reduction in power on FPGAs. ASICs experiment illustrate similar 48% reduction in circuit area, 27% delay reduction and 65% power reduction for a 14nm ASIC.
- 25 September 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 28 (1) , 115-128
AddNet: Deep Neural Networks Using FPGA-Optimized Multipliers
Low-precision arithmetic operations to accelerate deep-learning applications on field-programmable gate arrays (FPGAs) have been studied extensively, because they offer the potential to save silicon area or increase throughput. However, these benefits come at the cost of a decrease in accuracy. In this article, we demonstrate that reconfigurable constant coefficient multipliers (RCCMs) offer a better alternative for saving the silicon area than utilizing low-precision arithmetic. RCCMs multiply input values by a restricted choice of coefficients using only adders, subtractors, bit shifts, and multiplexers (MUXes), meaning that they can be heavily optimized for FPGAs. We propose a family of RCCMs tailored to FPGA logic elements to ensure their efficient utilization. To minimize information loss from quantization, we then develop novel training techniques that map the possible coefficient representations of the RCCMs to neural network weight parameter distributions. This enables the usage of the RCCMs in hardware, while maintaining high accuracy. We demonstrate the benefits of these techniques using AlexNet, ResNet-18, and ResNet-50 networks. The resulting implementations achieve up to 50% resource savings over traditional 8-bit quantized networks, translating to significant speedups and power savings. Our RCCM with the lowest resource requirements exceeds 6-bit fixed point accuracy, while all other implementations with RCCMs achieve at least similar accuracy to an 8-bit uniformly quantized design, while achieving significant resource savings.
- 25 September 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
- Vol. 28 (1) , 115-128
AddNet: Deep Neural Networks Using FPGA-Optimized Multipliers
Low-precision arithmetic operations to accelerate deep-learning applications on field-programmable gate arrays (FPGAs) have been studied extensively, because they offer the potential to save silicon area or increase throughput. However, these benefits come at the cost of a decrease in accuracy. In this article, we demonstrate that reconfigurable constant coefficient multipliers (RCCMs) offer a better alternative for saving the silicon area than utilizing low-precision arithmetic. RCCMs multiply input values by a restricted choice of coefficients using only adders, subtractors, bit shifts, and multiplexers (MUXes), meaning that they can be heavily optimized for FPGAs. We propose a family of RCCMs tailored to FPGA logic elements to ensure their efficient utilization. To minimize information loss from quantization, we then develop novel training techniques that map the possible coefficient representations of the RCCMs to neural network weight parameter distributions. This enables the usage of the RCCMs in hardware, while maintaining high accuracy. We demonstrate the benefits of these techniques using AlexNet, ResNet-18, and ResNet-50 networks. The resulting implementations achieve up to 50% resource savings over traditional 8-bit quantized networks, translating to significant speedups and power savings. Our RCCM with the lowest resource requirements exceeds 6-bit fixed point accuracy, while all other implementations with RCCMs achieve at least similar accuracy to an 8-bit uniformly quantized design, while achieving significant resource savings.