Towards an Effective Intrusion Detection Model Using Focal Loss Variational Autoencoder for Internet of Things (IoT).

As the range of security attacks increases across diverse network applications, intrusion detection systems are of central interest. Such detection systems are more crucial for the Internet of Things (IoT) due to the voluminous and sensitive data it produces. However, the real-world network produces imbalanced traffic including different and unknown attack types. Due to this imbalanced nature of network traffic, the traditional learning-based detection techniques suffer from lower overall detection performance, higher false-positive rate, and lower minority-class attack detection rates. To address the issue, we propose a novel deep generative-based model called Class-wise Focal Loss Variational AutoEncoder (CFLVAE) which overcomes the data imbalance problem by generating new samples for minority attack classes. Furthermore, we design an effective and cost-sensitive objective function called Class-wise Focal Loss (CFL) to train the traditional Variational AutoEncoder (VAE). The CFL objective function focuses on different minority class samples and scrutinizes high-level feature representation of observed data. This leads the VAE to generate more realistic, diverse, and quality intrusion data to create a well-balanced intrusion dataset. The balanced dataset results in improving the intrusion detection accuracy of learning-based classifiers. Therefore, a Deep Neural Network (DNN) classifier with a unique architecture is then trained using the balanced intrusion dataset to enhance the detection performance. Moreover, we utilize a challenging and highly imbalanced intrusion dataset called NSL-KDD to conduct an extensive experiment with the proposed model. The results demonstrate that the proposed CFLVAE with DNN (CFLVAE-DNN) model obtains promising performance in generating realistic new intrusion data samples and achieves superior intrusion detection performance. Additionally, the proposed CFLVAE-DNN model outperforms several state-of-the-art data generation and traditional intrusion detection methods. Specifically, the CFLVAE-DNN achieves 88.08% overall intrusion detection accuracy and 3.77% false positive rate. More significantly, it obtains the highest low-frequency attack detection rates for U2R (79.25%) and R2L (67.5%) against all the state-of-the-art algorithms.

Comparing Neuroplasticity Changes Between High and Low Frequency Gait Training in Subacute Stroke: Protocol for a Randomized, Single-Blinded, Controlled Study.

BACKGROUND: Walking recovery post stroke can be slow and incomplete. Determining effective stroke rehabilitation frequency requires the assessment of neuroplasticity changes. Neurobiological signals from electroencephalogram (EEG) can measure neuroplasticity through incremental changes of these signals after rehabilitation. However, changes seen with a different frequency of rehabilitation require further investigation. It is hypothesized that the association between the incremental changes from EEG signals and the improved functional outcome measure scores are greater in higher rehabilitation frequency, implying enhanced neuroplasticity changes. OBJECTIVE: The purpose of this study is to identify the changes in the neurobiological signals from EEG, to associate these with functional outcome measures scores, and to compare their associations in different therapy frequency for gait rehabilitation among subacute stroke individuals. METHODS: A randomized, single-blinded, controlled study among patients with subacute stroke will be conducted with two groups: an intervention group (IG) and a control group (CG). Each participant in the IG and CG will receive therapy sessions three times a week (high frequency) and once a week (low frequency), respectively, for a total of 12 consecutive weeks. Each session will last for an hour with strengthening, balance, and gait training. The main variables to be assessed are the 6-Minute Walk Test (6MWT), Motor Assessment Scale (MAS), Berg Balance Scale (BBS), Modified Barthel Index (MBI), and quantitative EEG indices in the form of delta to alpha ratio (DAR) and delta-plus-theta to alpha-plus-beta ratio (DTABR). These will be measured at preintervention (R0) and postintervention (R1). Key analyses are to determine the changes in the 6MWT, MAS, BBS, MBI, DAR, and DTABR at R0 and R1 for the CG and IG. The changes in the DAR and DTABR will be analyzed for association with the changes in the 6MWT, MAS, BBS, and MBI to measure neuroplasticity changes for both the CG and IG. RESULTS: We have recruited 18 participants so far. We expect to publish our results in early 2023. CONCLUSIONS: These associations are expected to be positive in both groups, with a higher correlation in the IG compared to the CG, reflecting enhanced neuroplasticity changes and objective evaluation on the dose-response relationship. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/27935.

Revisiting Spasticity After Stroke: Clustering Clinical Characteristics for Identifying At-Risk Individuals.

PURPOSE: To collectively identify the clinical characteristics determining the risk of developing spasticity after stroke. PATIENTS AND METHODS: A cross-sectional study was conducted at a single rehabilitation outpatient clinic from June to December 2019. Inclusion criteria were stroke duration of over four weeks, aged 18 years and above. Exclusion criteria were presence of concurrent conditions other than stroke that could also lead to spasticity. Recruited patients were divided into "Spasticity" and "No spasticity" groups. Univariate analysis was deployed to identify significant predictive spasticity factors between the two groups followed by a two-step clustering approach for determining group of characteristics that collectively contributes to the risk of developing spasticity in the "Spasticity" group. RESULTS: A total of 216 post-stroke participants were recruited. The duration after stroke (p < 0.001) and the absence of hemisensory loss (p = 0.042) were two significant factors in the "Spasticity" group revealed by the univariate analysis. From a total of 98 participants with spasticity, the largest cluster of individuals (40 patients, 40.8%) was those within less than 20 months after stroke with moderate stroke and absence of hemisensory loss, while the smallest cluster was those within less than 20 months after severe stroke and absence of hemisensory loss (21 patients, 21.4%). CONCLUSION: Analyzing collectively the significant factors of developing spasticity may have the potential to be more clinically relevant in a heterogeneous post-stroke population that may assist in the spasticity management and treatment.

A Survey on Underwater Wireless Sensor Networks: Requirements, Taxonomy, Recent Advances, and Open Research Challenges.

The domain of underwater wireless sensor networks (UWSNs) had received a lot of attention recently due to its significant advanced capabilities in the ocean surveillance, marine monitoring and application deployment for detecting underwater targets. However, the literature have not compiled the state-of-the-art along its direction to discover the recent advancements which were fuelled by the underwater sensor technologies. Hence, this paper offers the newest analysis on the available evidences by reviewing studies in the past five years on various aspects that support network activities and applications in UWSN environments. This work was motivated by the need for robust and flexible solutions that can satisfy the requirements for the rapid development of the underwater wireless sensor networks. This paper identifies the key requirements for achieving essential services as well as common platforms for UWSN. It also contributes a taxonomy of the critical elements in UWSNs by devising a classification on architectural elements, communications, routing protocol and standards, security, and applications of UWSNs. Finally, the major challenges that remain open are presented as a guide for future research directions.

Performance Evaluation and Validation of QCM (Query Control Mechanism) for QoS-Enabled Layered-Based Clustering for Reactive Flooding in the Internet of Things.

Internet of Things (IoT) facilitates a wide range of applications through sensor-based connected devices that require bandwidth and other network resources. Enhancement of efficient utilization of a heterogeneous IoT network is an open optimization problem that is mostly suffered by network flooding. Redundant, unwanted, and flooded queries are major causes of inefficient utilization of resources. Several query control mechanisms in the literature claimed to cater to the issues related to bandwidth, cost, and Quality of Service (QoS). This research article presented a statistical performance evaluation of different query control mechanisms that addressed minimization of energy consumption, energy cost and network flooding. Specifically, it evaluated the performance measure of Query Control Mechanism (QCM) for QoS-enabled layered-based clustering for reactive flooding in the Internet of Things. By statistical means, this study inferred the significant achievement of the QCM algorithm that outperformed the prevailing algorithms, i.e., Divide-and-Conquer (DnC), Service Level Agreements (SLA), and Hybrid Energy-aware Clustering Protocol for IoT (Hy-IoT) for identification and elimination of redundant flooding queries. The inferential analysis for performance evaluation of algorithms was measured in terms of three scenarios, i.e., energy consumption, delays and throughput with different intervals of traffic, malicious mote and malicious mote with realistic condition. It is evident from the results that the QCM algorithm outperforms the existing algorithms and the statistical probability value "P" < 0.05 indicates the performance of QCM is significant at the 95% confidence interval. Hence, it could be inferred from findings that the performance of the QCM algorithm was substantial as compared to that of other algorithms.

LCSS-Based Algorithm for Computing Multivariate Data Set Similarity: A Case Study of Real-Time WSN Data.

Multivariate data sets are common in various application areas, such as wireless sensor networks (WSNs) and DNA analysis. A robust mechanism is required to compute their similarity indexes regardless of the environment and problem domain. This study describes the usefulness of a non-metric-based approach (i.e., longest common subsequence) in computing similarity indexes. Several non-metric-based algorithms are available in the literature, the most robust and reliable one is the dynamic programming-based technique. However, dynamic programming-based techniques are considered inefficient, particularly in the context of multivariate data sets. Furthermore, the classical approaches are not powerful enough in scenarios with multivariate data sets, sensor data or when the similarity indexes are extremely high or low. To address this issue, we propose an efficient algorithm to measure the similarity indexes of multivariate data sets using a non-metric-based methodology. The proposed algorithm performs exceptionally well on numerous multivariate data sets compared with the classical dynamic programming-based algorithms. The performance of the algorithms is evaluated on the basis of several benchmark data sets and a dynamic multivariate data set, which is obtained from a WSN deployed in the Ghulam Ishaq Khan (GIK) Institute of Engineering Sciences and Technology. Our evaluation suggests that the proposed algorithm can be approximately 39.9% more efficient than its counterparts for various data sets in terms of computational time.

A Localization-Free Interference and Energy Holes Minimization Routing for Underwater Wireless Sensor Networks.

Interference and energy holes formation in underwater wireless sensor networks (UWSNs) threaten the reliable delivery of data packets from a source to a destination. Interference also causes inefficient utilization of the limited battery power of the sensor nodes in that more power is consumed in the retransmission of the lost packets. Energy holes are dead nodes close to the surface of water, and their early death interrupts data delivery even when the network has live nodes. This paper proposes a localization-free interference and energy holes minimization (LF-IEHM) routing protocol for UWSNs. The proposed algorithm overcomes interference during data packet forwarding by defining a unique packet holding time for every sensor node. The energy holes formation is mitigated by a variable transmission range of the sensor nodes. As compared to the conventional routing protocols, the proposed protocol does not require the localization information of the sensor nodes, which is cumbersome and difficult to obtain, as nodes change their positions with water currents. Simulation results show superior performance of the proposed scheme in terms of packets received at the final destination and end-to-end delay.

Feature-Based Retinal Image Registration Using D-Saddle Feature.

Retinal image registration is important to assist diagnosis and monitor retinal diseases, such as diabetic retinopathy and glaucoma. However, registering retinal images for various registration applications requires the detection and distribution of feature points on the low-quality region that consists of vessels of varying contrast and sizes. A recent feature detector known as Saddle detects feature points on vessels that are poorly distributed and densely positioned on strong contrast vessels. Therefore, we propose a multiresolution difference of Gaussian pyramid with Saddle detector (D-Saddle) to detect feature points on the low-quality region that consists of vessels with varying contrast and sizes. D-Saddle is tested on Fundus Image Registration (FIRE) Dataset that consists of 134 retinal image pairs. Experimental results show that D-Saddle successfully registered 43% of retinal image pairs with average registration accuracy of 2.329 pixels while a lower success rate is observed in other four state-of-the-art retinal image registration methods GDB-ICP (28%), Harris-PIIFD (4%), H-M (16%), and Saddle (16%). Furthermore, the registration accuracy of D-Saddle has the weakest correlation (Spearman) with the intensity uniformity metric among all methods. Finally, the paired t-test shows that D-Saddle significantly improved the overall registration accuracy of the original Saddle.