As compared to loss-based algorithms which have been predominantly used in TCP implementations, congestion-based congestion control called BBR has shown much better performance in resource-abundant modern communication links. However, for a high influx of TCP sessions on the core switch, clusters in high performance compute (HPC) nodes and data centers face resource constraints because of the orchestration and relocation of workload across the resource pool. This article discusses how to resolve this problem commonly known as TCP Incast, through efficient queue management of the bottleneck link and modifying the BBR algorithm. TCP Incast issue is analyzed for two efficient versions of congestion control i.e., BBR and CUBIC in a highly overloaded convergent switch of the cluster. It is noticeable that the queuing delay and buffer buildup are two important parameters in causing TCP Incast. Hence, the M/G/1/B queuing model is used when multiple TCP sessions are generating the network traffic and different buffer buildup scenarios are analyzed in the bottleneck node of HPC clusters. Based on the findings of our queuing analysis, we propose a modified BBR algorithm that introduces additional controls like Incast shaping to deal with queue build-up during TCP Incast. The effects of these adjustments on performance parameters like flow completion time and average throughput are found to be significant when compared with standard BBR and CUBIC implementations.
A sensor network is made up of many sensors deployed in different areas to be monitored. They communicate with each other through a wireless medium. The routing of collected data in the wireless network consumes most of the energy of the network. In the literature, several routing approaches have been proposed to conserve the energy at the sensor level and overcome the challenges inherent in its limitations. In this paper, we propose a new low-energy routing protocol for power grids sensors based on an unsupervised clustering approach. Our protocol equitably harnesses the energy of the selected cluster-head nodes and conserves the energy dissipated when routing the captured data at the Base Station (BS). The simulation results show that our protocol reduces the energy dissipation and prolongs the network lifetime.
Coronavirus constitutes a family of RNA viruses causing respiratory tract infections in both humans and birds. A mild disease appears like the common cold, and in other cases, causes Severe acute respiratory syndrome (SARS), Middle East respiratory syndrome (MERS), or COVID-19. As compared to COVID-19, SARS and MERS were limited to certain countries. On the other hand, COVID-19 was declared a pandemic by the World Health Organization on 11th March 2020. In this research, we perform the bibliometric assessment of Coronavirus research using the Scopus database. We studied 27824 articles written by 64903 researchers from 1951 till 20th June 2020, published in 3858 different sources. More than 65% of research appeared in the form of articles. More than 34% of publications appeared in 2020, coinciding with the appearance of COVID-19. This also resulted in a sharp increase in the average citation from 2.2 observed in 2019 to 14.5 seen in the year 2020. The USA is the most-cited country, followed by China. Nevertheless, Russia appears as the most-cited country per year. Wang Y writes the highest number of papers. The top source, according to the H-index, is the Journal of Virology. Similarly, most papers are published by researchers from Huazhong University of Science and Technology, China.
Entity matching is a very significant requirement in data management systems. It is a very challenging task to identify given two entities, which belong to the same real-world entity. It is crucial because people express the same entity in many different ways. For instance, a person expresses state California as “California” and others express as “CA”. A human can easily identify these two entities are same, where it is difficult for a machine. A human can\'t match similar entities among millions and millions of real-world entities. Hence, there is a need for a robust framework, which can match the entities by their attributes. Entity matching is a complex problem, which can be resolved with the help of heterogeneous machine learning algorithms. Previous studies on Entity Matching (EM) or Entity Resolution (ER) problem mainly emphasize on aiming various transformation techniques for matching their entities, NLP based matching and recently by Deep Learning techniques. However, all these solutions often fail to produce accuracy or scalability of matching requirements or ignore some of the internal constraints. Failing of EM, result in tedious and expensive rework and ruin business decisions. In this paper, we propose a solution called the “Deep Semantic Homophonic Synonymy” (DSHS) algorithm, which is a combination of rule-based, NLP, and Deep Learning techniques. It describes how to overcome the difficulties in entity matching by introducing the sematic approach to match the entity by rule-based preprocessing, NLP based enrichment, phonetical sound, different levels of abbreviations, spelling corrections and their prior or secondary forms. The result of this solution is significantly outperforming to all prior models existing in the literature of EM. This solution has six modules such as Preprocessing, Transformation, Translation, Word Embedding, Attribute Similarity, and Classification for training and deep learning for prediction. The entire solution gives high accuracy and scalable solution as compared to prior deep learning models present in EM.
In this study is introduced the application Non-cooperative game theory to define strategies that allow a balance of the profits of agricultural companies. In this sense, the scenerios plated to sale of products derived of earthworm. In this mode, the theory of non-cooperative games with infinite repetitions, strategies are proposed to allow the growers to make the best decisions and obtain profits within a balance for entrepreneurs. Likewise, a set of strategies are defined that guarantee the balance of the profits in infinite periods of time. In this sense, the growers in the different scenarios are focused to maximize the profits. Thus, due to the nature of the business, this cooperation does not occur, since in the different periods of time each company will be looking for best profit. According to results of the case of study the best option is to produce fertilizer and humus in the same proportion.
In this paper, a new key agreement protocol is presented. The protocol uses exponentiations of matrices over GF(2) to establish the key agreement in only single step of message exchange. Security analysis of the protocol shows that the shared secret key is indistinguishable from the random under Decisional Diffie-Hellman (DDH) Assumption for subgroup of matrices over GF(2) with prime order, and furthermore, the analysis shows that, unlike many other exponentiation based protocols, security of the protocol goes beyond the level of security provided by (DDH) Assumption and intractability of Discrete Logarithm Problem (DLP). Actually, security of the protocol completely transcends the reliance on the DLP in the sense that breaking the DLP does not mean breaking the protocol. Complexity of brute force attack on the protocol is equivalent to exhaustive search for the secret key. Analysis of the performance demonstrates that the protocol is applicable to real-time applications.
In this paper, a new key-agreement scheme is proposed and analyzed. In addition to being provably secure in the shared secret key indistinguishability model under Decisional Diffie-Hellman assumption for subgroup of matrices over GF(2) with prime order, which considered as basic security requirement, the scheme has an interesting feature; it uses exponentiations over cyclic group using hidden secret subgroup generator as a platform for the key exchange, whereby-unlike many other exponentiation based key exchange schemes- it transcends the reliance on intractability of Discrete Logarithm Problem in its security.