Transactions on Networks and Communications

Fault-Tolerant Placement of Additional Mesh Nodes in Rural Wireless Mesh Networks: A Minimum Steiner Tree Based Centre of Mass With Bounded Edge Length

2021-08-18T11:00:11+00:00

Wireless mesh networks are presented as an attractive solution to reduce the digital divide between rural and developed areas. In a multi-hop fashion, they can cover larger spaces. However, their planning is subject to many constraints including robustness. In fact, the failure of a node may result in the partitioning of the network. The robustness of the network is therefore achieved by carefully placing additional nodes. This work tackles the problem of additional nodes minimization when planning bi and tri-connectivity from a given network. We propose a vertex augmentation approach inspired by the placement of Steiner points. The idea is to incrementally determine cut vertices and bridges in the network and to carefully place additional nodes to ensure connectivity, bi and tri-connectivity. The approach relies on an algorithm using the centre of mass of the blocks derived after the partitioning of the network. The proposed approach has been compared to a modified version of a former approach based on the Minimum Steiner Tree. The different experiments carried out show the competitiveness of the proposed approach to connect, bi-connect, and tri-connect the wireless mesh networks.

Sensors and wrong values

2021-07-08T10:36:00+00:00

In the world of IoT and BigData, sensor based data collection is a really important domain. Using these tools it is possible to stow large amounts of data collection sensors in a factory or in nature in harsh environments. However, in order to obtain valuable information from these tools, it is important that potentially wrong data is discovered and handled. Automated exploration of wrong data is not a trivial task, even if similar measurements are performed in parallel with spatial differences. We present the difficulties of revealing defected data and suggest easy-to-implement procedures for detecting and handing them. We also draw attention to the potential disadvantages of these methods based on the given results.

Prediction of Breast Cancer images Classification Using Bidirectional Long Short Term Memory and Two-Dimensional Convolutional Neural network

2021-08-03T10:10:02+00:00

Breast cancer is most prevalent among women around the world and Nigeria is no exception in this menace. The increased in survival rate is due to the dramatic advancement in the screening methods, early diagnosis, and discovery in cancer treatments. There is an improvement in different strategies of breast cancer classification. A model for training deep neural networks for classiﬁcation of breast cancer in histopathological images was developed in this study. However, this images are aﬀected by data unbalance with the support of active learning. The output of the neural network on unlabeled samples was used to calculate weighted information entropy. It is utilized as uncertainty score for automatic selecting both samples with high and low conﬁdence. A threshold that decays over iteration number is used to decide which high conﬁdence samples should be concatenated with manually labeled samples and then used inﬁne-tuning of convolutional neural network. The neural network was optionally trained using weighted cross-entropy loss to better cope with bias towards the majority class. The developed model was compared with the existing model. The accuracy level of 98.3% was achieved for the developed model while the existing model 93.97%. The accuracy gain of 4.33%. was achieved as performance in the prediction of breast cancer .

Assessment of RCS-specific SNR and Loglikelihood Function in Detecting Low-observable Targets and Drones Illuminated by a Low Probability of Intercept Radar Operating in Littoral Regions

2021-07-05T08:36:25+00:00

The objective of this study is to deduce signal-to-noise ratio (SNR) based loglikelihood function involved in detecting low-observable targets (LoTs) including drones Illuminated by a low probability of intercept (LPI) radar operating in littoral regions. Detecting obscure targets and drones and tracking them in near-shore ambient require ascertaining signal-related track-scores determined as a function of radar cross section (RCS) of the target. The stochastic aspects of the RCS depend on non-kinetic features of radar echoes due to target-specific (geometry and material) characteristics; as well as, the associated radar signals signify randomly-implied, kinetic signatures inasmuch as, the spatial aspects of the targets fluctuate significantly as a result of random aspect-angle variations caused by self-maneuvering and/or by remote manipulations (as in drones). Hence, the resulting mean RCS value would decide the SNR and loglikelihood ratio (LR) of radar signals gathered from the echoes and relevant track-scores decide the performance capabilities of the radar. A specific study proposed here thereof refers to developing computationally- tractable algorithm(s) towards detecting and tracking hostile LoTs and/or drones flying at low altitudes over the sea (at a given range, R) in littoral regions by an LPI radar. Estimation of relevant detection-theoretic parameters and decide track-scores in terms of maximum likelihood (ML) estimates are presented and discussed.