Заседания научного семинара СТиМТТ в 2016-2017 г

10 March 2017 , 16:00, Str. Ordzhonikidze, 3, Rm. 110, 2nd floor

Title: «Bounds of probability characteristics for some non-stationary queueing systems»

Abstract:

Some classes of nonstationary queueing models are considered in this work, including model with batch arrivals and retrial queuing system. Intensities of arrivals and services are assumed to be locally integrable functions depend on t and on the current state of the system. We deal with the queue-length process and employ the general approach for  the study of  forward Kolmogorov system via logarithmic norm of an operator function and related estimates. We obtain bounds on the rate of convergence, truncations and perturbations for the considered queue-length processes. Specific queuing systems are studied.

Reporter: K. M. Kiseleva, Vologda State University.


11 November 2016 , 16:00, Str. Ordzhonikidze, 3, Rm. 110, 2nd floor. From 15:45 supposed video broadcast workshop meetings on the webinar of the department: http://webinar.sci.pfu.edu.ru:5080/

Title: «Queueing systems with LCFS pre-emptive generalized probabilistic priority and their application in performance evaluation of big-data batch processing systems»

Abstract:

The report focuses on the implementation of new queueing model in response time estimations for some batch processing systems. In the thesis consideration is given to a new type of M/G/1/r queuing system with LCFS pre-emptive generalized probabilistic priority policy (LCFS GPP). The main feature of this policy is that the service time of the newly arriving customer can be changed according to some probabilistic rules during system operation. There are known batch processing systems (e.g., Hadoop), the scheduler of which has to deal with the inaccurate job size information. Thus mathematical models, as well as simulation models, of such systems allow one to obtain only inaccurate estimations of the true mean response time. It will be shown numerically that in some cases the proposed approach allows one to obtain using mathematical modeling more accurate (compared to other scheduling policies like FCFS and PS) estimations of system’s mean response time.

Reporter: Lusine A. Meykhanadzhyan, PhD student at the department of applied probability and informatics of Peoples Friendship University of Russia. 

Презентация


21 October 2016 , 16:00, Str. Ordzhonikidze, 3, Rm. 110, 2nd floor. From 15:45 supposed video broadcast workshop meetings on the webinar of the department: http://webinar.sci.pfu.edu.ru:5080/

Title: Design and analysis of self-organization mechanisms aimed at ensuring the quality of service in mobile ad hoc networks

Abstract:

Self-organization is the key problem when designing mobile ad hoc networks. Self-organization mechanisms should adapt to the current state of the network, while maximizing its throughput and ensuring quality of service. The aim of the thesis is the design and analysis of the algorithms for network information dissemination, resource allocation and relay selection that ensure quality of service in mobile ad hoc networks. We develop an analytical model of the incremental network information dissemination mechanism that allows to estimate the probability that each network node has up to date information about all the other network nodes. We propose an original slots allocation algorithm for multi-hop transmission of voice streams with guaranteed quality of service and a new distributed relay selection algorithm for multicasting aimed at the formation of a minimum Steiner connected dominating set.

Reporter: Pavel Nekrasov, Moscow Institute of Physics and Technology (State University), Department for information transmission problems and data analysis.


7 October 2016 , 16:00 , Str. Ordzhonikidze, 3, Rm. 110, 2nd floor. From 15:45 supposed video broadcast workshop meetings on the webinar of the department:http://webinar.sci.pfu.edu.ru:5080/

Title: Development and Modeling of Methods Used in MAC Protocols for Real-time Video Streaming in Wi-Fi Networks

Abstract:

In recent years, we have witnessed a rapid growth of data transmitted over wireless networks. The major portion of such data is audio and video streams, among which particular attention should be paid to the so-called real-time streams (e.g., audio and video conferencing). Real-time streaming imposes stringent quality of service (QoS) requirements, which are typically expressed as the upper bound of Packet Loss Ratio (PLR) and maximum delay. Meeting QoS requirements in wireless networks, particularly in distributed Wi-Fi networks, is a difficult task.

For example, if packets destined to different stations are stored in a single FIFO queue at a Wi-Fi access point, even a short-term link failure for one of the recipients can lead to a significant increase in a packet service time for all other recipients, and thus can cause degradation of the quality of the received videos. To solve this problem, it is necessary to develop a packet queuing algorithm which takes into account the peculiarities of wireless networks, as well as QoS requirements.

Besides random noise in a wireless environment, transmission errors can be caused by packet collisions. Such a problem is becoming more complex and challenging due to the rapid growth in the number of stations in wireless networks, as well as in the number of overlapped wireless networks. A drastic solution to the problem is to use a deterministic channel access based on a channel reservation. However, even within the reserved time intervals the probability of a successful packet transmission attempt is less than 100% because of random noise and interference from distant hidden stations, and so packet retransmissions are required. Thus, it is important to find out the minimum amount of channel resource to be reserved for the transmission of a video stream with given QoS requirements, which is a nontrivial problem.

Reporter: Anton Kiryanov, researcher at IITP RAS, Ph.D. student at MIPT, the Department of Information Transmission Problems and Data Analysis in IITP RAS

Презентация


Get in touch with us

Контакты

Юридический адрес:

Россия, 117198,  Москва, ул. Миклухо-Маклая, д. 6, факультет физико-математических и естественных наук, кафедра прикладной информатики и теории вероятностей

Фактический адрес:

Россия, 115419, Москва, ул. Орджоникидзе, д. 3, факультет физико-математических и естественных наук, кафедра прикладной информатики и теории вероятностей, комн. 118

Тел.:  +7(495)952-28-23,

Факс: +7(495)952-28-23

 

Education - This is a contributing Drupal Theme
Design by WeebPal.