Seminar for
DECISION MAKING – THEORY, TECHNOLOGY AND PRACTICE
PROGRAM
Plan rada Seminara Odlučivanje - teorija, tehnologija, praksa za DECEMBAR 2021.
ČETVRTAK, 02.12.2021. u 13:00, Live stream Niš
Đorđe Jovanović, Mathematical Institute of the Serbian Academy of Science and Arts
EARLY DETECTION OF BOTNET INFECTED DEVICES USING NETWORK FLOW ANOMALY DETECTION
Botnets are a class of viruses whose defining characteristic is infecting a large number of devices, called bots, and controlling their behavior via Internet. The hacker who infected these devices, called a botmaster, then performs malicious activities using these bots. The two most common malicious activities performed by botnets are stealing bank credentials and distributed denial of service (DDoS) attacks. With the advent of Mirai and Gafgyt botnets, which target IoT devices, botmasters have been able to perform gigabyte-sized attacks, showing that botnets are becoming an increasing security issue. The first part of this talk will be focused on a brief overview of botnet detection. Historical and current examples of botnets, presently used datasets and research methodologies will be highlighted. The second part of this talk will introduce a novel approach to botnet detection, whose focus is on stopping Command and Control (CnC) communication early on, thus disabling the botnet. Using a Software Defined Network (SDN) controller to prevent malicious flows, this approach can be implemented as a real-time security solution. Finally, first results of botnet detection using machine learning models will be presented.
This is a joint work with Prof. dr Pavle Vuletić.
ČETVRTAK, 09.12.2021. u 13:00, Live stream Niš
Jelena Velimirović, Mathematical Institute of the Serbian Academy of Science and Arts
USING INFLUENCE DIAGRAMS WITH INTERVAL PROBABILITIES TO ASSESS RISK OF CIRCUIT BREAKERS' REPLACEMENT
This lecture deals with uncertainty and risk modelling in a complex power system. Influence diagrams with interval probabilities are chosen for the predictive modelling scenario of replacing the most critical circuit breakers in optimal time. Based on the available data on circuit breakers and other variables that affect the considered model of a complex power system, a group of experts was able to assess the situation using interval probabilities instead of crisp probabilities. Furthermore, this lecture will show how the confidence interval width affects decision-making in this context and eliminates the information asymmetry of different experts. The proposed approach clearly indicates the advantages of using interval probability when making decisions in systems such as the one considered in this lecture.
PONEDELJAK, 13.12.2021. u 13:00, Live stream Niš
Sanja Stevanović, Mathematical Institute of the Serbian Academy of Science and Arts
CONTROL VALUE IN SPACE SYNTAX
As basic mathematical structures used to visually describe relations within a set of objects, graphs had found numerous applications in many different sciences including architecture and urbanism, where these applications form a subfield named the space syntax. Graphs in space syntax are used to describe adjacency relations between different kinds of spaces in a building or a settlement, while their invariants then serve to attach numerical value to various properties of spaces. One such invariant is the control value vector CV, defined as $CV(u)=\sum_{v\in N(u)}\frac1{d(v)}$ where d(v) denotes the degree of a vertex v, while N(u) denotes the neighborhood of a vertex u in a given connected graph. This equation can be understood to model a process in which each vertex (space) shares a unit resource (such as access to the space) equally among its neighbors, and the control value of a vertex then represents the total amount of resources that the vertex obtains from its neighbors. In this talk we will present the basics of space syntax and discuss the use of the control value and the principal eigenvector for various types of spaces in residential buildings.
ČETVRTAK, 16.12.2021. u 13:00, Live stream Niš
Petar Vranić, Mathematical Institute of the Serbian Academy of Science and Arts
GISCIENCE AND APPLICATION OF GIS IN ENERGY SECTOR RESEARCH
The geospatial analytics market size is witnessing a historical increase in recent years emphasizing the growing importance of GIScience and Geographical Information Systems (GIS)role both in research and practice. GIS provides an excellent medium for data integration and a basis for a spatial decision-support system. This lecture will discuss GIS as one of the leading technologies in the spatial decision-support domain. The first part of the lecture will shed a light on opportunities for GIS development that comes with the introduction of new technologies like real-time data analytics, new mobile geospatial sensor platforms, etc., while the second part will present examples of GIS application in energy research.
PONEDELJAK, 20.12.2021. u 13:00, Live stream Niš
Dragan Urošević, Mathematical Institute of the Serbian Academy of Science and Arts
VARIABLE NEIGHBORHOOD SEARCH FOR CAPACITATED DISPERSION PROBLEM
The classical max min dispersion problem consists of selecting a certain number of elements (location, position) from a given set of elements so that the two closest selected elements are as far apart as possible. But, in the variant so called Capacitated dispersion problem, the elements are assigned capacities, so it is necessary to select the elements such that the total capacity of the selected elements is not less than the given limit value. The paper proposes a method based on variable neighborhood search. The method was implemented and comparisons were made on a large collection of instances with existing methods (tabu search method and two variants of scatter search).
ČETVRTAK, 23.12.2021. u 13:00, Live stream Niš
Lazar Velimirović, Mathematical Institute of the Serbian Academy of Science and Arts
APPLICATIONS OF ALGEBRAICALLY CONSTRUCTED SHORT SEQUENCE FAMILIES FOR 5G NON-ORTHOGONAL MULTIPLE ACCESS TECHNIQUES
Due to the substantial increase in the number of devices in 5G networks traditional multiple access techniques cannot utilize the time frequency space enough to satisfy the demand. One way to increase the utilization of the available time-frequency space is to apply non-orthogonal multiple access (NOMA) techniques coupled with complex message passing algorithms at the receiver are one way of increasing the user capacity in 5G networks. In NOMA users occupy a same resource element in any given time, and user separation is achieved through different interference cancellation strategies. For instance, in NOMA multi-user shared access (MUSA) techniques that use short spreading codes interference cancellation (IC), is necessary since because of their short length the codes do not have full orthogonality. In the other version of NOMA, pattern division multiple access (PDMA), sequences represent patterns whose elements are allocated same resource elements. Other multiple access techniques like Interleaved multiple access are also possible. In the paper we first give a method of an algebraic construction of short quasi orthogonal PDMA patterns that are ideally suited for use in NOMA PDMA. Not only there is a large number of patterns in the family (where we define a family as a set of patterns with the same length and weight), but they have excellent separation properties. The construction is based on properties of polynomials over finite fields. We show that within each family it is also possible to construct patterns with different weights (level of diversity). We then extend the construction to complex sequences and show that the sequences can also be used in MUSA. Using link level simulations, we establish the performance of our sequences, and compare them with pseudorandom noise (PN) sequences as well as with sequences in the other PDMA and MUSA related literature. The advantage of the proposed ways of constructing our sequences applied in IoT systems in relation to random sequences is reflected in the improve system performance.
Predavanja su namenjena sirokom krugu slusalaca, ukljucujuci studente redovnih i doktorskih studija. Seminar će se održavati svakog drugog četvrtka od 13:00 - 14:00h, CIITLAB, Elektronski fakultet Niš, Aleksandra Medvedeva 14, Niš
dr Lazar Velimirović
Rukovodilac seminara
dr Petar Vranić
Sekretar seminara
