A Tutorial on Federated Learning from Theory to Practice:Foundations,Software Frameworks,Exemplary Use Cases,and Selected Trends

When data privacy is imposed as a necessity,Federated learning(FL)emerges as a relevant artificial intelligence field for developing machine learning(ML)models in a distributed and decentralized environment.FL allows ML models to be trained on local devices without any need for centralized data transfer,thereby reducing both the exposure of sensitive data and the possibility of data interception by malicious third parties.This paradigm has gained momentum in the last few years,spurred by the plethora of real-world applications that have leveraged its ability to improve the efficiency of distributed learning and to accommodate numerous participants with their data sources.By virtue of FL,models can be learned from all such distributed data sources while preserving data privacy.The aim of this paper is to provide a practical tutorial on FL,including a short methodology and a systematic analysis of existing software frameworks.Furthermore,our tutorial provides exemplary cases of study from three complementary perspectives:i)Foundations of FL,describing the main components of FL,from key elements to FL categories;ii)Implementation guidelines and exemplary cases of study,by systematically examining the functionalities provided by existing software frameworks for FL deployment,devising a methodology to design a FL scenario,and providing exemplary cases of study with source code for different ML approaches;and iii)Trends,shortly reviewing a non-exhaustive list of research directions that are under active investigation in the current FL landscape.The ultimate purpose of this work is to establish itself as a referential work for researchers,developers,and data scientists willing to explore the capabilities of FL in practical applications.

Role of artificial intelligence in multidisciplinary imaging diagnosis of gastrointestinal diseases

The use of artificial intelligence-based tools is regarded as a promising approach to increase clinical efficiency in diagnostic imaging,improve the interpretability of results,and support decision-making for the detection and prevention of diseases.Radiology,endoscopy and pathology images are suitable for deep-learning analysis,potentially changing the way care is delivered in gastroenterology.The aim of this review is to examine the key aspects of different neural network architectures used for the evaluation of gastrointestinal conditions,by discussing how different models behave in critical tasks,such as lesion detection or characterization(i.e.the distinction between benign and malignant lesions of the esophagus,the stomach and the colon).To this end,we provide an overview on recent achievements and future prospects in deep learning methods applied to the analysis of radiology,endoscopy and histologic whole-slide images of the gastrointestinal tract.

Proactive Content Delivery for Vehicles over Cellular Networks:the Fundamental Benefits of Computing and Caching

The emergence of self-driving technologies implies that a future vehicle will likely become an entertainment center that demands personalized multimedia contents with very high quality. The surge of vehicular content demand brings significant challenges for the fifth generation(5G) cellular communication network. To cope with the challenge of massive content delivery, previous studies suggested that the 5G mobile edge network should be designed to integrate communication, computing, and cache(3C) resources to enable advanced functionalities such as proactive content delivery and in-network caching. However, the fundamental benefits achievable by computing and caching in mobile communications networks are not yet properly understood. This paper proposes a novel theoretical framework to characterize the tradeoff among computing, cache, and communication resources required by the mobile edge network to fulfill the task of content delivery. Analytical and numerical results are obtained to characterize the 3C resource tradeoff curve. These results reveal key insights into the fundamental benefits of computing and caching in vehicular mobile content delivery networks.

Apparent diffusion behaviour of intermolecular double-quantum coherence modulated by a distant dipolar field in solution NMR

A modified correlated spectroscopy （COSY） revamped with asymmetric Z-gradient echo detection sequence was designed to investigate the influence of diffusion hehaviour on intermolecular double-quantum coherence signal attenuation during the pre-acquisition period. Theoretical formulas were deduced and experimental measurements and simulations were performed. It is found that the diffusion behaviour of intermolecular double-quantum coherence in the pre-acquisition period may be different from that of conventional single-quantum coherence, depending on the relative orientation of diffusion weighting gradients to coherence selection gradients. When the orientation of the diffusion weighting gradients is parallel or anti-parallel to the orientation of the coherence selection gradients, the diffusion is modulated by the distant dipolar field. This study is helpful for understanding the signal properties in intermolecular double-quantum coherence magnetic resonance imaging.

Millimeter-Wave Absorption Properties of Thin Wave Absorber in Free Space with New Porous Carbon Material

In this paper, we focus on PHYTOPOROUS, a porous carbon material made entirely from plant-based ingredients, as a new broadband-wave absorber material that acts in the millimeter wave band. We created prototypes of thin rubber-sheet wave absorbers that contain porous carbon (PHYTOPOROUS) made from rice chaff and soybean hulls, which are both agricultural residue products that are generated in large quantities. We investigated the permittivity and reflectance characteristics of this material using the free-space time-domain method. The thin rubber-sheet wave absorber that contained PHYTOPOROUS made from soybean hulls exhibited a frequency band that was approximately 18 GHz wide and centered at 90 GHz. The return loss for this material was greater than −20 dB. This demonstrates that the material provides nearly constant reflection absorption over a wide frequency band.

Effect of Changes in Sea-Surface State on Statistical Characteristics of Sea Clutter with X-Band Radar

We have made observations of X-band radar sea clutter from the sea surface and sea-surface state in the Uraga Suido Traffic Route, which is used by ships entering and leaving Tokyo Bay, and the nearby Daini Kaiho Sea Fortress. We estimated the distributions of reflected amplitudes due to sea clutter using models that assume Weibull, Log-Weibull, Log-normal, and K-distributions. We then compared the results of estimating these distributions with sea-surface state data to investigate the effects of changes in the sea-surface state on the statistical characteristics of sea clutter. As a result, we showed that observed sub-ranges not containing a target conformed better to the Weibull distribution regardless of Significant Wave Height (SWH). Further, sub-ranges conforming to the Log-Weibull or Log-normal distribution in areas contained a target when the SWH was large, and as SWH decreases, sub-ranges conforming to a Log-normal. We also showed that for observed sub-ranges not containing a target, the shape parameter, c, of both Weibull and Log-Weibull distribution correlated with SWH. The correlation between wave period and shape parameters of Weibull and Log-Weibull distribution showed a weak correlation.

Time-domain training sequences design for MIMO OFDM channel estimation

This paper describes a Least Squares (LS) channel estimation scheme for MIMO OFDM systems based on time-domain training sequence. We first compute the minimum mean square error (MSE) of the LS channel estimation, and then derive the optimal criteria of the training sequence with respect to the minimum MSE. It is shown that optimal time-domain training sequence should satisfy two criteria. First, the autocorrelation of the sequence transmitted from the same antenna is an impulse function in a region longer than the channel maximum delay. Second, the cross-correlation between sequences transmitted from different antennas is zero in this region. Simulation results show that the estimator using optimal time-domain training sequences has better performance than that using optimal frequency training sequence at low signal-to-noise ratio (SNR). To reduce the training overhead, a suboptimal training sequence is also proposed. Comparing with optimal training sequence, it has low computation complexity and high transmission efficiency at the expense of little performance degradation.

Reference Selection for Offline Hybrid Siamese Signature Verification Systems

This paper presents an off-line handwritten signature verification system based on the Siamese network,where a hybrid architecture is used.The Residual neural Network(ResNet)is used to realize a powerful feature extraction model such that Writer Independent(WI)features can be effectively learned.A single-layer Siamese Neural Network(NN)is used to realize a Writer Dependent(WD)classifier such that the storage space can be minimized.For the purpose of reducing the impact of the high intraclass variability of the signature and ensuring that the Siamese network can learn more effectively,we propose a method of selecting a reference signature as one of the inputs for the Siamese network.To take full advantage of the reference signature,we modify the conventional contrastive loss function to enhance the accuracy.By using the proposed techniques,the accuracy of the system can be increased by 5.9%.Based on the GPDS signature dataset,the proposed system is able to achieve an accuracy of 94.61%which is better than the accuracy achieved by the current state-of-the-art work.

Resource Allocation Based on SFLA Algorithm for D2D Multicast Communications

Multicast device-to-device(D2D)communication technology is considered as one of the new technologies in the fifth generation(5G)networks that directly addresses the need for content sharing among internet users.In fact,when direct communication is available between devices,the spectral efficiency is improved by reusing the licensed cellular spectrum.The current studies show that D2D communication increases network capacity and reduces latency.In order to achieve the alternate capabilities,coordination is required to implement interference management.We considered subcarrier allocation for the uplink,in addition to the power control that takes place on the underlay network.The completed data rate in single multicast communication is significantly reduced and limited by nodes with lower channel quality.In this paper,we used Shuffled Frog Leaping Algorithm(SFLA)for resource allocation(RA)in D2D multicast communications.We compared the results of the SFLA algorithm with the Firefly Algorithm(FA),Ant Colony Optimization(ACO)and Particle Swarm Optimization(PSO);in terms of D2D user throughput,Cellular User(CU)throughput,network average throughput,network interference and signal interference noise ratio(SINR)target.The simulation results show that SFLA clearly outperforms other algorithms in terms of data rate under the high pressure of infeasibility.

Asymptotic sum rate of MMSE receivers in multi-user distributed antenna systems

In this paper, the asymptotic sum rate of a multi-user distributed antenna system （DAS） is analyzed. To mitigate inter-user interference, minimum mean squared error （MMSE） receivers are utilized to cooperatively process received signals in the uplink. It shows that inter-user interference is efficiently mitigated and the uplink sum rate of a multi-user DAS is greatly improved by adopting MMSE receivers. For very large number of users and remote antennas, the asymptotic uplink sum rate of MMSE receivers is derived by using virtue of the random matrix theory, which can be The approximation is verified to be quite accurate by Monte Carlo simply calculated in an iterative way simulations.

A Linear Array Antenna of Microstrip Patch Antennas Fed by the Open-End of Coplanar Waveguides

In this study, we constructed a 4-element linear array antenna using four 20 GHz band microstrip patch antennas with a structure such that the signal is fed to the patch antennas from open-end coplanar waveguides without contact. We investigated factors related to the design of linear array patch antennas. To adjust the maximum radiation direction and reduce return loss, we optimized the spacing between the elements and their shape. With an element spacing of 11.50 mm, patch width of 3.90 mm, and patch length of 4.15 mm, we obtained a resonance frequency of 20.05 GHz and a return loss of -29.59 dB at the resonance frequency. However, in the case of a 4-element linear array antenna structure, undesired resonances occurred in frequency bands other than the design resonance frequency band of 20 GHz. To suppress these undesired resonances and obtaining stable operation at the design frequency, we propose a new structure in which the feed line is loaded with a short stub, and show through computer simulations that the occurrence of undesired resonances can be sufficiently suppressed. Furthermore, we demonstrate the problem of radiation gain reduction caused by introducing a short stub, propose a design method for a new structure in which the feed line has slits between the stubs, and show improvement of the antenna gain by 0.5 dBi.

Advances in Digital Front-End and Software RF Processing: PartⅡ

In the first editorial of this two-part special issue, we pointed out that one of the biggest trends in wireless broadband, radar, sonar, and broadcasting technology is software RF processing and digital front-end [1]. Thistrend encompasses signal processing algorithms and integrated circuit design and includes digital pre-distortion （DPD）, conversions between digital and analog signals, digita up-conversion （DUC）, digital down-conversion （DDC）, DC offset,

Developments and Applications of Data Deidentification Technology under Big Data

In this age characterized by rapid growth in the volume of data,data deidentification technologies have become crucial in facilitating the analysis of sensitive information.For instance,healthcare information must be processed through deidentification procedures before being passed to data analysis agencies in order to prevent any exposure of personal details that would violate privacy.As such,privacy protection issues associated with the release of data and data mining have become a popular field of study in the domain of big data.As a strict and verifiable definition of privacy,differential privacy has attracted noteworthy attention and widespread research in recent years.In this study,we analyze the advantages of differential privacy protection mechanisms in comparison to traditional deidentification data protection methods.Furthermore,we examine and analyze the basic theories of differential privacy and relevant studies regarding data release and data mining.

Performance Analysis of the D-STTD Communication System with AMC Scheme

In this paper, we propose a Double-Space Time Transmit Diversity (D-STTD) communication system with Adaptive Modulation and Coding (AMC) scheme and analyze its performance using simulation experiments. The simulation results show that the probability of selecting a high Modulation and Coding Scheme (MCS) level increased as the Signal to Noise Ratio (SNR) improved. Furthermore, the D-STTD communication system with AMC scheme provided a more uniform throughput distribution throughout the entire SNR range compared to its counterpart which did not apply AMC scheme. Also, the maximum throughput of the D-STTD communication system with AMC scheme was twice that of a conventional AMC communication system or a Space Time Transmit Diversity (STTD) communication system with AMC scheme.

Suppression of Log-Normal Distributed Weather Clutter Observed by an S-Band Radar

We have observed weather clutter containing targets (ships) using an S-band radar with a frequency 3.05 GHz, a beam width 1.8°, and a pulsewidth 0.5 μs. To investigate the weather clutter amplitude statistics, we introduce the Akaike Information Criterion (AIC). We have found that the weather clutter amplitudes obey the log-normal, Weibull, and log-Weibull distributions with the shape parameters of 0.308 to 0.470, 4.42 to 4.51, and 15.91 to 16.44, respectively, for small data within the beam width of an antenna. We have proposed the log-normal/CFAR circuit modified a Cell-Averaging (CA) LOG/CFAR circuit. It is found that weather clutter is suppressed with improvement of 51.58 dB by log-normal/CFAR. As a result, we have showed that weather clutter observed by S-band radar does not obey the Rayleigh distribution and our log-normal/CFAR circuit has an effect on suppression of clutter and detection of target, while conventional LOG/CFAR circuit does not. In addition, if our circuit can be realized, we will have an advantage economically.

An efficient priority service model with two-level-polling scheme

This paper considers an efficient priority service model with two-level-polling scheme which the message packets conform to the discrete-time Geom/G/1 queue with multiple vacations and bulk arrival. By the embedded Markov chain theory and the probability generating function method, we set up the mathematics functions and give closed form expressions for obtaining the mean cyclic period （MCP）, the mean queue length （MQL） and the mean waiting time （MWT） characteristics, the analytical results are also verified through extensive computer simulations. The performance analysis reveals that this priority polling scheme can gives better efficiency as well as impartiality in terms of system characteristics, and it can be used for differentiating priority service to guarantee better QoS and system stability in design and improvement of MAC protocol.

License Plate Recognition for Parking Control System by Mathematical Morphology

Nowadays, license plate recognition for parking systems is a critical task to provide automatic control of customers and payment. This paper introduces a new method for automatic recognition of license plates of vehicles by mathematical morphology. The proposed method can provide the license plate number of the plates in different light conditions, colors, sizes, and inclination （angles）. The algorithm can recognize the license plates of European Union vehicles quickly and correctly. The pattern learning of mathematical skeletons has high efficiency in the process. The performance of the algorithm is demonstrated well by the test in a parking control system.

Accurate Analysis on Bluetooth Low Energy Neighbor Discovery

The basic concept of Bluetooth Low Energy (BLE) is short packet transmission and transient connection. It can quickly establish a connection, send data, and quickly disconnect, so that neighbor discovery is frequent and becomes an important issue. In the neighbor discovery which includes advertising and scanning, the BLE specification defines several important parameters. The parameters on the advertiser side include advertising interval, advertising duration, etc. On the scanner side, there are scan interval, scan window, etc. How to configure these parameters for quick neighbor discovery has been troublesome for BLE implementers. Prior analyses on BLE discovery process also showed some disagreements or made some incorrect assumptions. In this paper, we use rigorous probability-theory based derivations to obtain different kinds of successful discovery probabilities. We clarify disagreements in prior works and also provide insights on how to configure parameters for maximizing discovery probability. In particular, we prove that the discovery probabilities on each of the three channels are correlated. We also find that, when the advertising duration is set close to some multiples of the scan interval, an ill-fated synchronization problem will occur. To have a high discovery probability, both scan window and scan interval should be set at a large value, though it might not be good for energy saving.

Equal Preference Multi-Path Routing for L2 Hierarchical Networks

The layer 2 network technology is extending beyond its traditional local area implementation and finding wider acceptance in provider’s metropolitan area networks and large-scale cloud data center networks. This is mainly due to its plug-and-play capability and native mobility support. Many efforts have been put to increase the bisection bandwidth in a layer 2 network, which has been constrained by the spanning tree protocol that a layer 2 network uses for preventing looping. The recent trend is to incorporate layer 3’s routing approach into a layer 2 network so that multiple paths can be used for forwarding traffic between any source-destination (S-D) node pair. ECMP (equal cost multipath) is one such example. However, ECMP may still be limited in generating multiple paths due to its shortest path (lowest cost) requirement. In this paper, we consider a non-shortest-path routing approach, called EPMP (Equal Preference Multi-Path) that can generate more paths than ECMP. The EPMP is based on the ordered semi-group algebra. In the EPMP routing, paths that differ in traditionally-defined costs, such as hops, bandwidth, etc., can be made equally preferred and thus become candidate paths. We found that, in comparison with ECMP, EPMP routing not only generates more paths, provides higher bisection bandwidth, but also allows bottleneck links in a hierarchical network to be identified when different traffic patterns are applied. EPMP is also versatile in that it can use various ways of path preference calculations to control the number and the length of paths, making it flexible (like policy-based routing) but also objective (like shortest path first routing) in calculating preferred paths.