Search

Nowadays, WiFi is mostly operating in 2.4 GHz band, which allows only 3 non-overlapping channels. For deployment in building it could be a problem to repeat only three channels. In this paper we compare standard three channel scheme (channels 1, 6 and 11) with four channel schemes. The most promising scheme is with channels 1, 5, 9 and 13, which involves overlapping channels, but the overlap is minimal and in certain distances (power levels) is possible to use this scheme. The problem could be in some countries which do not allow usage of 13 channels. We also compare other schemes, with overlapping channels, but they have a much lower throughput than scheme with channels 1, 5, 9 and 13.

Software Defined Networking (SDN) is a new approach in design, implementation and maintenance of computer networks. With SDN, it is possible to dramatically reduce business costs because the whole network can be maintained in a centralized fashion which leads to a simplified and effective network management. This paper is concerned with SDN from educational point of view. The paper’s primary goal is to present beginning of a learning path to start experimenting with SDN programming and traffic path definition based on Hewlett-Packard (HP) SDN solution. Useful sources of documentation were selected and two experiments were performed: firstly an already existing HP SDN application was extended in the area of network visualization, secondly and experiment of traffic path definition with real network switches was performed.

This article deals with the impact of Wireless (Wi-Fi) networks on the perceived quality of voice services. The Quality of Service (QoS) metrics must be monitored in the computer network during the voice data transmission to ensure proper voice service quality the end-user has paid for, especially in the wireless networks. In addition to the QoS, research area called Quality of Experience (QoE) provides metrics and methods for quality evaluation from the end-user’s perspective. This article focuses on a QoE estimation of Voice over IP (VoIP) calls in the wireless networks using network simulator. Results contribute to voice quality estimation based on characteristics of the wireless network and location of a wireless client.

A management of computer classroom is undoubtedly a difficult task for the administrator which has to prepare virtual operating systems for education. It is quite common that lectors need to edit the particular machine during the semester, and that is the case where the main problems can appear. The process of changes deployment is not just very time-consuming but during it a virtual machine inconsistency can appear. The main part of this paper focuses on system process diagrams and its pseudocode. At first, the machine is created on the remote server by lector or administrator. After a proper approval, the machine is able to be deployed. The lector then specifies the details about date, time and destinations of the virtual machine deployment. Once these details are approved, the virtual machine will be automatically deployed at the specified time. The automatic deployment includes also an initial configuration of the virtual machine at remote desktop and its post-install configuration (hostname, MAC address, etc.). Once all steps are completed, the process will be marked as succeed. We present an automatized solution which provides a possibility how to easily manage computer classroom with virtual operating systems. The proposed solution should deliver a greater flexibility, more reliability and faster deployment in comparison with the current solution used in our computer classroom. The proposal is also able to manipulate with already deployed machines for easy changes (e.g. software updates). The main advantage is the improvement of classroom management process automation.

In this paper we inspect a difference between original sound recording and signal captured after streaming this original recording over a network loaded with a heavy traffic. There are several kinds of failures occurring in the captured recording caused by network congestion. We try to find a method how to evaluate correctness of streamed audio. Usually there are metrics based on a human perception of a signal such as “signal is clear, without audible failures”, “signal is having some failures but it is understandable”, or “signal is inarticulate”. These approaches need to be statistically evaluated on a broad set of respondents, which is time and resource consuming. We try to propose some metrics based on signal properties allowing us to compare the original and captured recording. We use algorithm called Dynamic Time Warping (Müller, 2007) commonly used for time series comparison in this paper. Some other time series exploration approaches can be found in (Fejfar, 2011) and (Fejfar, 2012). The data was acquired in our network laboratory simulating network traffic by downloading files, streaming audio and video simultaneously. Our former experiment inspected Quality of Service (QoS) and its impact on failures of received audio data stream. This experiment is focused on the comparison of sound recordings rather than network mechanism.We focus, in this paper, on a real time audio stream such as a telephone call, where it is not possible to stream audio in advance to a “pool”. Instead it is necessary to achieve as small delay as possible (between speaker voice recording and listener voice replay). We are using RTP protocol for streaming audio.

Computer network used for laboratory training and for different types of network and security experiments represents a special environment where hazardous activities take place, which may not affect any production system or network. It is common that students need to have administrator privileges in this case which makes the overall security and maintenance of such a network a difficult task. We present our solution which has proved its usability for more than three years. First of all, four user requirements on the laboratory network are defined (access to educational network devices, to laboratory services, to the Internet, and administrator privileges of the end hosts), and four essential security rules are stipulated (enforceable end host security, controlled network access, level of network access according to the user privilege level, and rules for hazardous experiments), which protect the rest of the laboratory infrastructure as well as the outer university network and the Internet. The main part of the paper is dedicated to a design and implementation of these usability and security rules. We present a physical diagram of a typical laboratory network based on multiple circuits connecting end hosts to different networks, and a layout of rack devices. After that, a topological diagram of the network is described which is based on different VLANs and port-based access control using the IEEE 802.1x/EAP-TLS/RADIUS authentication to achieve defined level of network access. In the second part of the paper, the latest innovation of our network is presented that covers a transition to the system virtualization at the end host devices – inspiration came from a similar solution deployed at the Department of Telecommunications at Brno University of Technology. This improvement enables a greater flexibility in the end hosts maintenance and a simultaneous network access to the educational devices as well as to the Internet. In the end, a vision of a system of virtual machines preparation and automated deployment tailored for our needs is briefly outlined.

In the area of economical classification tasks, the accuracy maximization is often used to evaluate classifier performance. Accuracy maximization (or error rate minimization) suffers from the assumption of equal false positive and false negative error costs. Furthermore, accuracy is not able to express true classifier performance under skewed class distribution. Due to these limitations, the use of accuracy on real tasks is questionable. In a real binary classification task, the difference between the costs of false positive and false negative error is usually critical. To overcome this issue, the Receiver Ope­rating Characteristic (ROC) method in relation to decision-analytic principles can be used. One essential advantage of this method is the possibility of classifier performance visualization by means of a ROC graph. This paper presents concrete examples of binary classification, where the inadequacy of accuracy as the evaluation metric is shown, and on the same examples the ROC method is applied. From the set of possible classification models, the probabilistic classifier with continuous output is under consideration. Mainly two questions are solved. Firstly, the selection of the best classifier from a set of possible classifiers. For example, accuracy metric rates two classifiers almost equiva­lently (87.7 % and 89.3 %), whereas decision analysis (via costs minimization) or ROC analysis reveal differe­nt performance according to target conditions of unequal error costs of false positives and false negatives. Secondly, the setting of an optimal decision threshold at classifier’s output. For example, accuracy maximization finds the optimal threshold at classifier’s output in value of 0.597, but the optimal threshold respecting higher costs of false negatives is discovered by costs minimization or ROC analysis in a value substantially lower (0.477).

The topic of this article is focused on problems related to enterprise financial supervising. In the concrete, the situation of enterprise investment policy evaluation is described here. In this case, as a convenient tool for decision support, the approach of artificial intelligence was selected, particularly the model of neuron network. For the purpose of enterprise economic state evaluation, we use four input variables which describe the economic state. Three main variables are selected and the fourth one is the additional. The coding of main variables is chosen with the respect to the possible states of the enterprise. The multilayer neuron network was used for evaluation.The neural network can solve problems, which are hardly solvable for a manager because there can exist a lot of factors affecting the final decision. We have to take into account the fact that sometimes the situation is too complex. In this case, when the system gives incorrect result, it is possible to extend the current learning set and add adequate patterns which will help the system to recognize states of the enterprise.