Continuous improvements in very-large-scale integration (VLSI) technology and design software have significantly broadened the scope of digital signal processing (DSP) applications. The use of application-specific integrated circuits (ASICs) and programmable digital signal processors for many DSP applications have changed, even though new system implementations based on reconfigurable computing are becoming more complex. Adaptable platforms that combine hardware and software programmability efficiency are rapidly maturing with discrete wavelet transformation (DWT) and sophisticated computerized design techniques, which are much needed in today’s modern world. New research and commercial efforts to sustain power optimization, cost savings, and improved runtime effectiveness have been initiated as initial reconfigurable technologies have emerged. Hence, in this paper, it is proposed that the DWT method can be implemented on a field-programmable gate array in a digital architecture (FPGA-DA). We examined the effects of quantization on DWT performance in classification problems to demonstrate its reliability concerning fixed-point math implementations. The Advanced Encryption Standard (AES) algorithm for DWT learning used in this architecture is less responsive to resampling errors than the previously proposed solution in the literature using the artificial neural networks (ANN) method. By reducing hardware area by 57%, the proposed system has a higher throughput rate of 88.72%, reliability analysis of 95.5% compared to the other standard methods.
Cryptographic algorithms can be used for security services that require low cost and minimal power consumption in a wide range of settings. Some examples of such technologies include wireless local area networks (WLAN), wireless personal area networks (WPAN), wireless sensor networks (WSN), and security tokens [
Compared to high-performance computing, FPGAs can achieve speeds of magnitude faster where integrated resources on FPGA are increasing [
FPGA is a general-purpose electronic component that can be programmed or reprogrammed by the designer after it has been installed in a system [
Digital signal processors that can be programmed in software are commonly used to accomplish a shared goal of system adaptability [
Reprogram ability and rapid construction times, and reduced effort in comparison with full-custom VLSI design, are some of the reasons to choose FPGA-based technology. In addition, the microelectronics production process has advanced to the point where FPGA-based digital systems can be designed with performance close to that of comprehensive layouts.
The major contribution of this paper is given as follows:
FPGA-DA is implemented to address the implementation issue using DWT learning. Digital hardware architecture has achieved high data throughput by partitioning repeated AES modules. Digital architecture for the DWT process is discussed through the implementation of FPGA. The proposed FPGA-DA highlights the novelty by its performance in hardware and software programmability efficiency with improved security using AES algorithm. This work further evaluated the ANN process in FGPA-DA to ensure the customizability by halving the number of iterations and computing all rotational directions in advance for reducing FPGA-DA area and computation time.
The remaining section of this paper is organized as: Section 2 provides a literature study on existing FPGA methods, whereas Section 3 describes the implementation of field-programmable gate array using a digital architecture (FPGA-DA) proposed in this research. The findings and conclusions of the experiments are presented in Sections 4, and 5 which conclude with our future scope of this research.
Programmable devices, such as programmable logic arrays (PLAs), have been used with programmable array logic, where devices began to be used as logic components, yet they suffered from a power consumption problem for several years. There are numerous methods for calculating the real number of a quantity in VLSI, each with its own set of considerations. However, using multiple algorithms to arrive at a precise result is difficult. Minimum energy dissipation in mixed-signal integrated circuits is required to store energy in reasonable battery size. To make a connection between the analog and digital logic worlds, the analog-to-digital converter is essential are compared with our proposed system mentioned below:
Many image and signal processing algorithms rely on the matrix multiplication kernel operation. Here, describing how to design and implement matrix multiplier configurations for image and sensor processing applications using field-programmable gate arrays. Image processing applications use dense matrix-vector multiplication, which is used in the first design. The Virtex-4 FPGA (V-FPGA) was used to implement the design, and the execution time on the FPGA was measured to assess its performance mentioned in [
An FPGA-based signal processing card is discussed in this paper. FPGAs are used to design an onboard real-time digital signal processing system. The platform can simultaneously decode a variety of technological and analog signals. This card’s design trend is compact size, high interconnection and fast real-time preparation. An FPGA-based signal processing card (FPGA-SPC) has been implemented [
Fixpoint one-dimensional (1-D) wavelet decomposition computation using lifting scheme (1-D WDC-LS) was proposed in this paper as a design scheme for area accurate and scalable speed pipeline VLSI architecture. The scheme’s primary goal is to reduce the number and period of clock cycles and the efficient area while using hardware resources as sparingly as possible, which is given [
In the end-user, communication systems, health sciences, and industrial markets, Co-ordinate Rotation Digital Computer-based channel estimation (CORDIC-CE) has become a critical tool, providing developers with the significant incentive for algorithm porting into architecture [
This paper describes the implementation of a hoist wavelet processor on a field-programmable gate array (FPGA) device for signal detection. Using an unsigned integer Haar lifting wavelet transform (UIHLWT), this processor implements our proposed algorithm for detecting target portions of signals [
An FPGA can be used to build an entire system in many cases, and this is an excellent option for applications that do not necessitate the effectiveness of custom hardware. The combination of FPGA logic blocks and inter-connects matrices, as well as one or more microelectronics, has led to new architectures that are more efficient in our proposed work FPGA-DA when compared with our other traditional methods.
Given that DSPs operate sequentially and cannot be parallelized, speed is reduced by the clock frequency of the DSPs in terms of performance. However, if an appropriate multithreaded architecture is designed for an FPGA, it can run quickly. Irrespective size of the executable programme, a DSP consumes power based on the number of memory elements it uses. In FPGAs, the amount of power consumed is determined by the design of the circuit. To incorporate a parallel algorithm, multiple components work together to implement the system’s functionality with FPGAs. The implementation of the proposed system FPGA-DA is discussed in the following.
The methodological framework for developing the hardware electronic design digitization (EDD) tools, methods, and FPGA technology works together to produce an optimized circuit for the final application. The performance of the design process can be significantly improved by using a combination of FPGA hardware, custom-designed IP cores, and EDA tools. In FPGA-DA, using the term design methodology refers to the FPGA design process as a whole. Algorithms are implemented in hardware design methodology. FPGA vendors use a variety of design flows, and they all follow a similar sequence of tasks are illustrated in the below
These are the most common and essential steps in the FPGA design process, starting with source code. One of the most important factors in creating an optimized digital circuit with FPGA is EDD tools. FPGA devices are used to implement the method or circuit described by data Input in this flow. Based on the level of complexity of the design, there are two standard approaches to specifying FPGA designs: HDL and schematic-based. FPGA designers primarily use high-definition language (HDL) design entry for more complex and computationally demanding algorithms, such as technology mapping. Once the design has been specified in HDLs or logic synthesis, the designer must verify logically correct. Simulated functional and behavioral is used to accomplish this by technical mapping using functional simulation. In general, FPGA designs placement is critical. This directly impacts the routing technique and effectiveness of a design on an FPGA. The placement algorithms can be grouped under the term routing with timing-driven. It is the final step in the design process before producing the bitstream to program the FPGA. Prefabricated resources like wire segments and multiplexers are required for FPGA routing, which is time-consuming.
For the next two iterations, transition and simple arithmetic operations are applied to the scale factor from read-only memory (ROM) to obtain the component at a time
FPGA delays are considered during timing simulations to ensure that the design is logically correct. The final step in the FPGA proposed design is to generate a bit stream and download the resulting bit file.
FPGA-based signal processing methods that work with both signal processing signals. Using (fast Fourier transform) FFT, this card provides the ability to speed up and test real-time process control design concepts. The new system image analysis kit dramatically reduces computation costs and facilitates the overall design process, making it more efficient.
After initializing all input and output devices on an FPGA-DA given in
The above-stated
Conclusions drawn from this paper show that FPGAs have many advantages over general-purpose microcontrollers in real-time applications, such as in this paper’s FPGA-DA system where VLSI signals and high-speed peripherals are used.
For the design and efficient implementation of an FPGA, where resources are extremely limited, matrix multiplication has been particularly challenging. Three performance metrics are commonly used to evaluate FPGA-based designs: frequency (transmission delay), location, and energy. FPGA fixed-point implementations are a popular choice because of their speed and low power consumption. Furthermore, a fixed-point matrix multiplier unit frequently requires less silicon in an FPGA or ASIC than its suspended counterpart.
A set of low
When the low pass and high pass filters are down analyzed, their outputs contain the low-frequency signals of the original signal, which are called approximate parts, and the high-frequency components, which are highly textured parts like edges of the original signal, which are mentioned in detail from the above
Latency and portion have been used as performance indicators for FPGA designs. Energy efficiency and power dissipation have become increasingly important as portable mobile devices have increased in the past few years. Several components in FPGA devices use significant amounts of power, and the configurable connections are responsible for a large portion of this power consumption.
Precisely defined dilations and interpretations of the analyzer’s function, known as a wavelet, represent a signal in the discrete wavelet transform. Generally, basis functions are used in the most common form of the DWT, which uses a relational grid. From the wavelet function, researchers can derive the following equations for the generation of wavelets:
From the equation as mentioned above
Two patterns dominate the ordering of the coefficients after filtering: one that acts as a regularization term like a time series, and the other that helps bring out the complete details of the data. In the AES algorithm, the DWT is computed at multiple configurations using a space representation convolution for low-pass
The VLSI based signal processing has been filled with a weighted average
Based on the above
Additionally, the control signal is a synthetic test message with additional noise. Filtering parameters can be calculated based on the various noise power and use the thresholding methodology on the selected features because it is starting to use that the vast majority of the noise is encapsulated within them. Using the soft thresholding method called widespread threshold calculated using below equation,
There are indeed additional noises
In the wavelet transform field, nonlinear edge detection of detail coefficients is called wavelet compression performed to identify. The wavelet transform is used to improve the image quality of the signal by dividing it into a small number of coefficients and then applying an adaptive threshold procedure to each one. Finally, all these are summed to get a synthesized output.
For an accurate assessment of the filtering abilities, different analysis functions reflected different, and levels of noise
As a result of the nonlinear transformation, it is possible to perform nonlinear sector is known on the training data by simply using a linear higher dimensional space to separate the points in the training dataset. VLSI can use nonlinear mappings unique mathematical relationships to as advantages, and the algorithm is explained below,
1:
2:
3: set t=
4:
5:
6: S
7: then
8:
9: if
10:
11:
12:
13:
This methodology can regulate learning machine complexity, including through a structural mathematical point of view from the above algorithm (1). The inputs are one of the reasons for AES success. In a statistical framework, this principle can regulate the training machine’s complexity and define the upper limits of its generalization ability. As an alternative, our design employs AES stages in addition to the ten stages of the AES itself to produce transmission stages with clock signal
Programmable digital signal processors are used to meet a common goal of system adaptability. Even though these platforms allow for flexible implementation thanks to programming interfaces and significant savings in performance and power efficiency, application developers and compilers must adapt their processing approach to the computing resources available on these platforms.
It can be seen from the
An AES algorithm based on training instances is used to calculate the core’s signal processing method. FPGA-DA outcomes can be tested with the MATLAB. The suggested FGPA-DA with DWT and AES system prototype was built on a virtex 6 FPGA with the XC6VCX75T package using Xilinx tools. However, it is an experimental method on the topic. It can be used to understand its primary properties, such as the number of clock cycles required to arrive at a workable solution, clock frequency, and device utilization. To achieve this goal, here focus on creating an RBF-DWT prototype and that the same process can be easily applied to other DWTs.
Factors | Metrics |
---|---|
Data provider | 250 |
Frequency maximum | 18.276 (MHz) |
Gate count | 148,215 |
Capacity | 250 MB |
Simulation time | T = 1,150,970 ns |
Bandwidth | 512 Mbps |
This setup is used to test the proposed system’s FPGA-DA efficacy under real-world conditions, using the parameters optimization comparison, cost savings, improved runtime effectiveness rate, throughput rate, and reliability analysis as shown in
Number of signals | V-FPGA | CORDIC-CE | FPGA-SPC | UIHLWT | FPGA-DA |
---|---|---|---|---|---|
10 | 53.7 | 63.7 | 75.7 | 85.4 | 91.7 |
20 | 54.8 | 65.8 | 76.8 | 85.5 | 92.8 |
30 | 55.7 | 67.7 | 77.7 | 86.7 | 92.7 |
40 | 56.7 | 68.7 | 78.7 | 87.9 | 93.7 |
50 | 57.9 | 69.9 | 79.9 | 88.3 | 93.9 |
60 | 58.8 | 70.8 | 80.8 | 88.1 | 94.8 |
70 | 60.6 | 71.6 | 81.6 | 89.3 | 94.6 |
80 | 61.8 | 72.8 | 82.8 | 90.9 | 95.8 |
90 | 62.5 | 73.0 | 83.9 | 90.5 | 95.1 |
100 | 63.0 | 74.6 | 84.9 | 90.5 | 96.3 |
Number of signals | V-FPGA | CORDIC-CE | FPGA-SPC | UIHLWT | FPGA-DA |
---|---|---|---|---|---|
10 | 44.4 | 52.3 | 68.7 | 66.8 | 78.4 |
20 | 46.7 | 53.8 | 81.5 | 74.5 | 76.8 |
30 | 45.9 | 54.2 | 85.6 | 79.8 | 74.8 |
40 | 47.8 | 56.7 | 75.8 | 81.5 | 81.2 |
50 | 51.4 | 55.9 | 76.8 | 83.4 | 83.4 |
60 | 53.2 | 57.2 | 71.5 | 82.4 | 87.84 |
70 | 54.1 | 60.4 | 73.5 | 88.5 | 72.85 |
80 | 55.8 | 62.3 | 81.5 | 84.2 | 82.54 |
90 | 60.08 | 62.5 | 76.8 | 86.7 | 86.57 |
100 | 62.2 | 63.0 | 83.5 | 81.4 | 88.72 |
Many important aspects to consider, such as alternatives, movement, and cost assessments, are part of the problem-solving space when solving multi-objective optimization problems. In FPGA-DA designing an ASIC is a way to reduce the cost and size of an integrated device, product, or system by integrating individual elements and their features into a single element is calculated using optimizing parameter mentioned in
It is possible to have a digital signal or a sustained signal because the domain of a continuous signal is time, e.g., a linked timeframe of the reals. The term prolonged signal refers to a constant amplitude and time obtained using
In our proposed system, regardless of the processing delay, samples that are being analyzed (input) and generated (output) continuously can be processed (or generated) as part of an actual digital signal computation process are calculated in
Although throughput is typically measured in bits per second (bps), it can be measured in data packets per second (p/sec) or time slots mentioned in
Based on
By halving the number of iterations and computing all rotational directions in advance, here present architecture of FPGA-DA that reduces both its surface area and computation time. The proposed architecture is highly configurable and extendable to higher precision as well. Comparison of proposed FPGA-DA with other unfolded architectures in the literature shows that the number of stages is reduced while unwanted signals are eliminated.
Over other traditional approaches, new hardware architecture for the AES algorithm has been proposed as (FPGA-DA) and compared to two AES hardware structures, iterative repetition and transmission approach. Using an FPGA, researchers can implement the structure in the real world. Here there is increased throughput by approximately 88.72%. Furthermore, it has a superior throughput advantage of 90.2% over the iterative looping AES structure; however, the area is 3.5 times larger than the sequential looping structure. Rather than evaluating 16 bytes of data simultaneously with all of the hardware requirements, the reordering of bytes streamlines the evaluation in parallel. Partitioning the AES into sub-blocks with intermediate buffers in between is another important feature, as it creates an extensive parallel processing structure for all AES blocks. DWT-based learning on a digital architecture has been discussed in this paper and is an example of how it is implemented on an FPGA device. In FPGA-DA, it is possible to improve the results obtained significantly. Because of this, slightly modified versions may exist that originate initialized operating at higher clock frequency bands. In the spirit of immediately preceding neural processors, future work will solve this challenge by integrating a general-purpose baseband processor and a learning controller module on the same chip.
The authors extend their appreciation to King Saud University for funding this work through Researchers Supporting Project number (RSP-2021/387), King Saud University, Riyadh, Saudi Arabia.