There are many ways of managing traffic in Info Communication networks. The priority ones are the most relevant for consideration to ensure the proper quality for services that have requirements for minimum delays or channel bandwidth. However, choosing the most appropriate way to manage traffic is a task that requires a comprehensive analytical approach. Within the universal package network, the telephony service is one of many services provided. In the corporate sector, there has been a noticeable increase in interest in IP telephony services in recent years, stimulated by some obvious advantages in the form of flexibility and openness of IP systems, the possibility of using various communication solutions (data, video, voice) within a single platform, as well as extensive use of wireless technologies. The purpose of the article is to study the effectiveness of traffic prioritization for Quality of Service (QoS) management in computer networks. Within the scope of the article, simulation modeling is carried out in AnyLogic, the behavior of packet delays of different classes is studied when using various service disciplines. The objects of the study are heterogeneous traffic and its prioritization disciplines with dynamic priorities, as well as Info - Communication systems and communication channels. The subject of the study is the disciplines of traffic maintenance with dynamic priorities within the framework of Info - Communication systems.
Today, the era of Big Data has come in the field of Info -Communications. The Internet affects almost all spheres of modern life. Every minute, a huge amount of structured and unstructured data is transmitted over global networks. There is also a stable annual growth in the number of Internet users.
The main problem of the Big Data era is to optimize resources and increase the efficiency of data
transmission. It should be noted that traffic cannot be considered as a single whole, for various network
services, certain types of network traffic should be selected that meet the efficiency criteria for the
maximum number of simultaneously working users. At the same time, practical limitations in the form
of the width of communication channels and the limited power of network equipment are taken into
account [
The work is intended to evaluate the effectiveness of traffic management mechanisms on the global
2021 Copyright for this paper by its authors. Comparative analysis of existing traffic management mechanisms In the article, the concept of quality of Service is considered, which implies the solution of two main tasks: 1. Creating and maintaining the order of receipt of packages 2. Minimizing delays and ensuring positive dynamics of packages transmission
The use of data package switching technology can not guarantee high throughput in information
and communication networks. The reason for this is the lack of guarantees for the delivery of the
package. In some applications, the order and delivery intervals do not affect the performance and quality
of user interaction. At the same time, for others, these parameters are fundamental. High traffic service
requirements are not met by the TCP and UDP transport layer protocols because TCP allows some
possible delays in delivery, although it guarantees the correct delivery of packets, and UDP can reduce
delays, but does not provide high-quality traffic service and mechanisms for implementing such. At the
same time, it is necessary to guarantee the delivery of such information like audio, video, and
multimedia in real-time with the minimum possible delay [
The quality of service is defined as "the total effect of the operating parameters of the service, which
determines the degree of user satisfaction with this service". (ITU-T Recommendation E. 800) To be
able to quantify the quality of service in the network, it is necessary to introduce some numerical
parameters. The following parameters are used to evaluate QoS:
Average packet Delivery Delay (IPPacket Transfer Delay). IPTD is defined as the sum of the
delivery times of all packets between the source and the recipient, divided by the number of packets
[
FirstInFirstOut is one of the simplest maintenance disciplines, the essence of which is to process packets in the same order in which they are initially queued. The use of this discipline in the case of processing large traffic flows simultaneously leads to the dominance of several of them, which negatively affects the efficiency of using network resources.
Priority Queueing is a maintenance discipline that involves using a combination of several queues
that are processed using the Taildrop or RandomEarlyDetection buffering disciplines and using the
FirstInFirstOut service discipline within themselves [
WeightedFairQueueing is a maintenance discipline in which a separate queue using the
FirstInFirstOut discipline is allocated for each traffic class, as well as the allocation of a certain share
of the channel bandwidth for each of these queues [
The capture of VoIP traffic is carried out as follows: the detection of opened Skype program ports, detection of the port and address that has the greatest activity during the call, the capture of the traffic with the filter (Fig. 1).
VOD traffic is captured as follows: determining the address through which video traffic goes using the browser's debugging console; capturing traffic with a filter (Fig. 2).
If we consider the packet length distribution functions for UDP and TCP traffic, we can see that
UDP packets (Fig. 3) have a fixed maximum size, unlike TCP packets (Fig. 4). In case of loss of a TCP
packet, it will be requested again, and the information will not be lost, unlike UDP - in case of loss of
a packet, it is lost forever[
FIFO - an elementary queue without prioritization: each traffic class receives the same amount of service, taking into account the delay when issuing in the communication channel. Experiments with throughput values did not lead to obtaining characteristics that meet the specified requirements （Fig. 5, Table 2). This result is associated with a large packet size and a small-time interval between packets: the specified buffer size and bandwidth are not enough for the selected type of traffic. When trying to increase the initial parameters, the following results were obtained (Fig. 6, Table 3)
Picture 6 shows that now the average stay time and the probability of losses are within the normal range. As a result of the iterative increase of throughput, the model met the QoS criteria better and better: the delay and the probability of loss decreased. Priority is set for different traffic classes: low-priority class traffic is transmitted only when there are no high-priority class packets in the queue. Thus, the best quality of service is provided for the highpriority class, but the low-priority class is blocked during overloads. It can be seen from the schemes (Fig. 7,8,9) and tables (Table 4,5) that the change in the bandwidth strongly affects low-priority traffic. As in the previous case, with variations of the bandwidth, it is not possible to achieve the required qualities. 4.3.
Weight is set for each traffic. For each cycle of the WFQ operation, packets are transmitted from the
queue of one class, the total size of which is equal to the weight of the class. Setting the weight
guarantees that a class with a higher weight will receive a higher quality of service, and that in
conditions of high load, the class will receive a channel in a finite time[
When the weights were varied, the results improved slightly (Pic. 10.11). As in the previous case, with the variation of the bandwidth, it is not possible to achieve the required qualities. The lowest bandwidth value was achieved at 23000 bps (Table 6,7). When varying the weights at this throughput and some values below, it was not possible to achieve significantly better results.
The WFQ discipline gives zero latency for high-priority packets, unlike PQ. In both cases, changing the bandwidth greatly affects low-priority traffic. (Fig. 12,13)
As a result of the work, the following results were obtained: 1. When the bandwidth of the channel increases, the characteristics decrease to a certain threshold, after which the bandwidth increase has no effect 2. The FIFO Service Discipline does not provide traffic management mechanisms. 3. The PQ Service Discipline provides an elementary control mechanism that cannot handle overload cases. 4. The WFQ Service Discipline provides a traffic management mechanism by assigning weights to the classes into which traffic is divided; it is more flexible than the previous two, and it copes with congestion better than PQ. 5. The initial configuration (6 Mbit/s bandwidth and 600 Kbyte/s buffer) is unacceptable when applying any Service Discipline in the case of simultaneous use of Skype and Twitch. 6. If it is necessary to separate the quality of service and the minimum bandwidth, WFQ will be a suitable option, otherwise, FIFO will do.