This work proposes an Internet of Medical Things (IoMT) framework to offer a low-cost, portable, and effective solution for early cervical cancer screening through an intelligent integration of autofluorescence (AF) imaging microscopy, cloud server and edge computing as software-defined network (SDN) and deep learning (DL) techniques. Wavelet transform, on red, green and blue channel of AF images are done followed by the fusion techniques that highlight various cytological details and to mitigate data scarcity problem autoencoder based Generative Adversarial Network (GAN) is employed. Finally, two DL classifier models of Convolutional Neural Network (CNN) architectures, namely VGG16 and ResNet50 are trained for cervical cancer detection on the fused images at SDN controller along with other computation extensive image processing operations. Once the DL network is trained, a pre-trained submodel is run for AF test images on edge node accessed by patients’ smart phones (mobile apps). A large set of results show that along with low latency and energy consumption, ResNet50 model achieves an accuracy of 0.99, outperforming the existing methods.

Reconfigurable intelligent surface (RIS) is considered a promising technology in sixth-generation (6G) networks due to its ability to configure the phase of signals using low-cost reflecting elements. An RIS-aided device-to-device (D2D) communication system operating in underlay mode is considered in this work. Assuming the presence of a direct link between the D2D pair, we present a closed-form expression of signal-to-interference and noise ratio (SINR) at the D2D user. We aim to solve the SINR maximisation problem by jointly optimising the power allocated to the D2D source and the RIS placement. A closed-form global optimal solution to the proposed optimisation problem has been obtained. Numerical results are presented to validate the analytical framework. Our proposed solution can yield a significant gain of around 4.65 dB over the benchmark.

This work considers a cognitive radio (CR) network consisting of a set of CR transmit-receive node pairs, one fusion center (FC), multiple primary user emulation attack (PUEA) nodes, an eavesdropper ($E_{av}$) node and a set of friendly jammers used for protection of CR data transmission from eavesdropping. At the initial time slot of the frame, simultaneous energy harvesting (EH) and spectrum sensing are done through power splitting (PS) mode. CR transmit nodes then amplify and forward the sensed samples of both the primary user (PU) and the PUEA to the FC for cooperative spectrum sensing (CSS). Based on the CSS decision, CR transmit nodes either perform EH over the entire duration or make an opportunistic data transmission in time division mode. The closed form expressions of the optimal sensing duration, power allocation factor and transmit power for each secondary user (SU) are found. The sum secondary throughput of the network is maximized under the constraints of meeting the sensing reliability of the PU, individual energy causality for each SU and the best selected jammer, interference at the PU receiver, individual secondary and secrecy outage probability. Simulation results show a performance gain on the maximum value of the sum secondary throughput by$\sim$17.56 and$\sim$49.69 percent over the existing works.

Abstract Compressed sensing (CS) or compressive sampling has shown an enormous potential to reconstruct a signal from its highly under‐sampled observations. A high dimensional image processing system can adopt the CS paradigm to reduce the storage and the transmission burden. However, a large sensing system (matrix) is required to capture high dimensional images. The CS reconstruction algorithms demand heavy computational requirement due to the use of large sensing matrix. It sometimes becomes impractical to implement a sensing system of the desired size due to the limited access to the CPU memory. To address this issue, the present work proposes a deep learning based CS framework that uses a convolutional neural network to enable capturing of a high dimensional image by utilizing a multi‐layer subsampling and filtering operations. The proposed approach uses another convolutional neural network that reconstructs the original image without depending on the sensing network at a significantly reduced computational cost. Extensive simulation results show that the proposed method reconstructs a high dimensional image with an improved PSNR value by dB, SSIM value by and FSIM value by compared to the other state‐of‐the‐art methods in less than 1.4 seconds.

Recent progress in cognitive radio networks (CRNs) promises to meet device‐to‐device communication requirements for spectrum utilisation and power control to support billions of machines/devices to be connected worldwide. The architecture of the CRN must maintain a high data rate (throughput) at low power consumption, which requires both spectrum efficient and energy efficient system design. To this aim, the proposed work adopts a CRN model, which operates in an interweave mode that allows spectrum sensing followed by opportunistic secondary user (SU) data transmission over the unused bandwidth of the primary user (PU) in a non‐overlapping frame structure. Closed form expressions of the optimal spectrum sensing duration, bandwidth, and power allocation of each secondary node are obtained to maximise the sum throughput of the overall CRN while maintaining the constraints of sensing reliability of PU, individual SU transmission outage probability, permissible interference, and residual bandwidth. Numerical results highlight that the proposed scheme maximises the sum network throughput by and over the other existing techniques.

Cooperation among the sensing nodes in cognitive radio (CR) networks improves spectrum sensing (SS) reliability. However, determining the efficient set of secondary users (SUs) to be involved in cooperative SS (CSS) becomes challenging in vehicular networks (VNs) due to the mobility of primary user (PU) and/or SU nodes. This work explores an energy detection‐based CSS scheme in CR enabled VNs (CR‐VNs) considering the uniform velocity motion of PU and SU nodes. A distance‐dependent distribution function is developed first to find the probability of a SU resides in a specific coverage of the PU transmission zone. This distribution function is then used in deriving the expression of the probability of detection and probability of false alarm using majority rule at the fusion centre. The efficacy of the mathematical analysis is studied through a large set of simulation results that highlight the enhanced SS performance in terms of the probability of detection by , and at uniform velocity over the existing works in CR‐VN.

Massive multiple‐input–multiple‐output (MIMO) emerges as a promising technology to meet the ever growing demand on a large volume of data transmission in an energy efficient manner in the upcoming fifth generation networks. This explores an integrated system model of improved wireless channel estimation (CE) and its consequent effect on image reconstruction, both done in compressed sensing (CS) framework. A massive MIMO channel exhibits sparse nature and has been modelled as a CS problem. Two CS‐MIMO‐CE algorithms, discrete cosine transform‐based stagewise orthogonal matching pursuit and improved stagewise orthogonal matching pursuit (iStOMP), are suggested here. Low density parity check codes are also integrated to serve three‐fold purposes, as a pilot signal, sensing matrix, and error control code, while Daubechies transform offers a sparse representation of the MIMO channel and transmitted image. The proposed iStOMP MIMO‐CE algorithm offers gain in CE over the existing works and gain in 2 dB on image visual quality at 75% CS measurements.

This work considers an energy harvesting (EH) based two‐hop relay assisted cognitive radio system that consists of a secondary transmitter, a relay node and a cooperative jammer operated in an underlay mode with a primary transmitter–receiver pair. An optimisation problem is formulated to minimise the secrecy outage probability under the constraints of simultaneously meeting the energy causality of the secondary transmit nodes, interference to the primary receiver and secondary outage probability. An analytical expression of the secrecy outage probability is developed followed by deriving the closed‐form expressions to determine the optimal transmit power of the source, the relay node and the fraction of the time slot for EH. Simulation results show that the minimum value of the secrecy outage probability is and less over the existing works.

This study explores neovascularisation and lesion detection in an integrated framework for gradation in diabetic retinopathy (DR). Imaging is assumed to be done from sub‐sample measurements following compressed sensing. Blind estimation of the scale of the matched filter (MF) followed by fuzzy entropy maximisation is done for extraction and classification of the thick and the thin vessels. Mutual information (MI) between vessel density and tortuosity of the thin vessel class is maximised in two dimensions (2D) for neovascularisation detection. For lesion detection, MI between the maximum MF response and the maximum Laplacian of Gaussian filter response is jointly maximised in 2D. The outcomes are then combined in a common platform for gradation in DR. Simulation results demonstrate that 95% images of each of DRIVE, STARE and DIARETDB1 databases and 94% images of MESSIDOR database are correctly graded by the proposed method when 80% measurement space is considered.

An adaptive plateau limit‐based histogram equalisation algorithm is suggested to enhance digital images. Histogram of the image is clipped with a plateau limit to avoid over enhancement. The plateau limit is derived from the average of the mean and the median intensity values to offer the improved enhancement. Clipped histogram is subdivided into three parts, using histogram subdivision limit parameters that are calculated on the basis of the standard deviation of the image. Histogram of individual sub‐image is equalised independently and then combined into a single enhanced image. Experimental results demonstrate that the proposed plateau limit‐based tri‐histogram equalisation algorithm enhances the image quality. Compared with the other traditional plateau and non‐plateau limit‐based histogram equalisation algorithms, quantitative and visual quality assessments effectively validate the superiority of the proposed algorithm.

Recent advancement in wireless communications demands high data rate transmission at low power consumption. To this aim, the present work proposes an energy harvesting based cognitive radio (CR) system operating in time slotted mode. Energy harvesting and spectrum sensing (SS) are done simultaneously by the CR user through power splitting mode. On the other hand, during the transmission time slot, the CR user either harvests energy or transmits its own data based on SS decision result. An optimization problem is developed to maximize the residual energy under the constraints of sensing reliability of the PU and spectrum efficiency (SE) target rate of the CR. The concavity of the objective function is shown which ensures a global solution to the problem. Closed form expressions for the sensing time and CR transmit power are found and the results are validated by simulations.

This paper proposes a new method based on multiple thresholds for automatic extraction of blood vessels specially from a low contrast and non-uniformly illuminated background of retina. Curvelet transform is used to extract the finest details along the vessels since it can represent the lines, the edges and the curvatures very well. Next matched filtering is done to intensify the blood vessels' response in the enhanced image. The multiple threshold values for the maximum matched filter response that maximize the fuzzy entropy are considered to be the optimal thresholds to extract the different types of vessel silhouettes from the background. Differential Evolution algorithm is used to specify the optimal combination of the fuzzy parameters. Performance is evaluated on publicly available DRIVE database and is compared with the existing blood vessel extraction methods. Simulation results demonstrate that the proposed method outperforms the existing methods in detecting the long and the thick as well as the short and the thin vessels.

This paper addresses the automatic blood vessel detection problem in retinal images using matched filtering in an integrated system design platform that involves curve let transform and fuzzy c-means. Although noise is kept constant in medical CCD cameras, due to a number of factors, the contrast between the background and the blood vessels in retinal images and consequently the visual quality of the images looks very poor. Some form of pre-processing operation is therefore essential for the accurate extraction of these blood vessels. Since curve let transform can represent lines, edges and curvatures very well as compared to other multi-resolution techniques, this paper uses curve let transform to enhance the retinal vasculature. Matched filtering is then used to intensify the blood vessels which is further employed by fuzzy c-means algorithm to extract the vessel silhouette from the background. Performance is evaluated on publicly available DRIVE database and is compared with the existing blood vessel extraction methodology that uses curve let transform. Simulation results demonstrate that the proposed method is very much efficient in detecting long and thick as well as short and thin vessels, wherein the existing methods fail to extract tiny and thin vessels.

This paper proposes a joint secret sharing and data hiding scheme to serve the purpose of progressive quality access control on digital images. A gray scale secret image is divided into noise-like shares using bit plane modulation. Individual share is then embedded into respective cover images (from a set of images) in DCT domain using m-ary spread spectrum modulation. This joint data hiding and secret sharing scheme is analogous to as (k, n) scheme in secret sharing; however progressive quality access control of the secret is possible to achieve based on the particular group of users involved in decoding process. Robustness of the proposed method is also studied against a set of common signal processing operations. Extensive simulation results are also shown to validate the efficacy of the proposed method.

Reversible watermarking (RW) is used to embed some useful information into digital content in a reversible manner, where the reconstruction of the original image is highly desirable. This reversibility is important for almost all sensitive image application domains such as medical and military fields. To this aim here we have proposed a hybrid algorithm that offers higher embedding bit rate with a considerable low visual distortion. The proposed algorithm is a hybrid model of modified reversible contrast mapping (RCM) and prediction error expansion (PEE) used in RW. The benefits of RCM that provides higher embedding space and PEE algorithm that allows higher embedding bit rate are used combinedly. Both the attributes are used here in one integrated form that allows high data embedding rate with low embedding distortion. In a single pass embedding, the proposed algorithm provides 1.33 bpp, 1.15 bpp, 1.024 bpp and 1.085 bpp for Lena, Boat, Baboon and Elaine images, respectively.

Segmentation of medical images is a very difficult and challenging task due to many inherent complex characteristics present in it. Moreover in many practical situations, medical images are captured at low measurement spaces i.e. at compressed sensing (CS) paradigm for a variety of reasons, for example, due to the limited number of sensors used or measurements may be extremely expensive. Reconstructed medical images after CS operation are found to have uneven intensity values as well as blurred non-uniform shape of the organs. Although conventional discrete wavelets are widely used for edge enhancement and detection, but may not be efficient for detecting the curvatures of the different small organs. Curvelet transform, which is a multi-scale multiresolution transform, can be used as pre-processing for enhancement prior to segmentation of medical images rich with curvatures and missing or broken boundaries. In the proposed work, first reconstruction of a Magnetic Resonance (MR) image is done at multi channel CS platform using a weighted fusion rule. Curvelet transform is then applied on reconstructed MR images to obtain detailed image by suppressing the approximate subband. A sharpen image is developed which is then used for segmentation based on calculating a set of thresholds by maximizing fuzzy entropy for different fuzzy functions. Parameters for the different fuzzy functions that indicate memberships of the different regions, are determined by the Genetic Algorithms (GAs). Extensive simulation results are shown to highlight the performance improvement by the proposed method.

This paper proposes an optimal algorithm that minimizes watermark power (hence embedding distortion) subject to meeting the strict data hiding rate constraint in presence of joint additive noise and random gain attack. The proposed algorithm approaches such problem using convex optimization framework and gets the solution for watermark power and host sample allocation. This offers an optimality in terms of embedding distortion-robustness-data hiding capacity with polynomial computation complexity. Simulation carried over convolution coded integer wavelet coefficients on compressed host image show that ~ 4.5 to ~ 3 times less normalized watermark power is required in the proposed system. This leads to an improvement of ~ 10 dB in document- to-watermark ratio over direct embedding on entropy coded data. Simulation results also show that an improvement in bit error rate of 10-2is achieved over fixed memoryless attack channel.

This paper proposes a data hiding scheme to serve dual purpose of quality access control and error concealment of digital images. This is accomplished by projecting host image first on N-mutually orthogonal sample sets followed by embedding of an encoded binary watermark (host digest) using quantization index modulation (QIM) but without complete self-noise suppression. It is well known that due to insertion of external information, there would be degradation in visual quality of the host image. This has been used here to play the key role in access control through reversible process and error concealment using the extracted image digest. Decision variable for each bit of watermark decoding is formed from the weighted average of N-decision statistics that increases the correctness of extracted watermark bits. This improved detection enables self-noise suppression by authorized user, error concealment in fading radio mobile channel and thus leads to avail better quality image. Simulation results have shown the validity of our claims along with relative comparative results for other existing works.

This paper proposes an improved detection algorithm for spread spectrum watermark through the realization of an efficient group combined interference cancellation (GCIC) technique. First, each watermark bit is spread over N-mutually distinct points of the cover signal and the decision variable at the decoder is formed from the weighted average of N-decision statistics. Attack channel is modeled as Rayleigh distribution followed by AWGN (additive white Gaussian noise). Diversity technique is then applied to form the resultant watermarked image from the weighted average of multiple copies in order to improve detection performance where WSIR (watermark signal to interference ratio) is used as weight factor to achieve low BER (bit error rate). Genetic algorithms (GA) is used to partition embedded bits into multiple groups, namely very strong, strong, weak and very weak based on the normalized magnitudes of their decision variables. Performance of the GA based multiple group combined interference cancellation (MGCIC) for four, three and two groups with conventional interference cancellation (IC) are then reported. A significant improvement in detection performance and capacity is seen with GA based group combined interference cancellation (GCIC) relative to conventional interference cancellation (IC) at the cost of slight increase in computation complexity. Simulation results show that bit error rate (BER) for watermark decoding is reduced as the embedded bits are divided into more number of groups and optimal partitioning of these groups are made through the use of GA.

This paper proposes a new model of high capacity carrier interferometry/multicarrier code division multiple access (CI/MCCDMA) system through the simultaneous support of high and low data rate transmission using odd and even subcarriers. Reduction in peak-to-average power ratio (PAPR) is achieved through the phase shift of even (odd) CI code using odd (even) sub-carriers by an amount of ¿/2 and odd (even) CI code using odd (even) sub-carriers by -¿/2, all angles measured with respect to the orthogonal codes supporting high data rate transmission. The CI code optimization is done for further reduction in PAPR values. Specifically, assuming a frequency selective Rayleigh-fading channel, bit error rate (BER) at receiver is improved through the cancelation of MAI (multiple access interference) at subcarriers level using PIC (parallel interference cance-lation) at reduced computation cost. Finally, the system is optimized with respect to the number of subcarriers, number of users (capacity) and signal-to-noise ratio (SNR) using GAs (Genetic Algorithms) to achieve an acceptable set of values for BER, PAPR and average data rate (ADR).

In this letter, a new model of spread spectrum (SS) watermarking is proposed where each watermark bit is spread using a distinct code pattern over N-mutually orthogonal signal points. Decision variable for each bit of watermark decoding is formed from the weighted average of N-decision statistics. Multiple detection stages are applied where the detected bit patterns of one stage are used in the immediate subsequent stage to cancel the multiple bit interference (MBI) effect experienced by the individual bit. Fuzzy logic is used to beat the effect of such MBI through Multiple Group Combined Interference Cancelation (MGCIC) technique in the intermediate stage(s). The group is formed from the decision magnitudes corresponding to a group of code patterns and the ensemble of code patterns are classified in different groups like very strong, strong, weak and very weak based on fuzzy membership function derived from the magnitude of the decision variables. Attack channel is modeled as Rayleigh distribution followed by AWGN (additive white Gaussian noise). Simulation results show that BER (bit error rate) value is reduced more, of course, at the cost of slight increase in computational complexity if the code patterns are classified in a greater number of groups.