Abstract on the topic: "Efficiency analysis of packet transmission scheduling algorithms in LTE networks"
Mobile networks of the fourth generation, based on the use of multiple access technologies with orthogonal modulation OFDMA and the method of spatial coding of the MIMO signal, make it possible to significantly increase the transmission of traffic from subscribers.
The beginning of the 21st century is associated with a radical change in transport telecommunication networks - mobile communications have become the main type of communication in the world. By the end of the first decade, mobile communications and wireless access began to be seen as interacting radio access networks Radio Access Network. GSM EDGE RAN (GERAN), UMTS Terrestrial RAN (UTRAN), Wireless Local Area Network (Wi-Fi) and other 3rd generation standards are radio access networks [1] .
The distinctive features of these networks are: architecture flexibility, the ability to dynamically change the network topology when connecting, moving and disconnecting mobile subscribers, high data transfer rate, a high degree of protection against unauthorized access, as well as the rejection of expensive and not always possible installation or lease fiber optic or copper cable. In LTE networks, control functions were transferred to base stations, which, in addition to servicing the radio part, began to make decisions about the routing of subscriber traffic. At the same time, one of the main problems is the problem of traffic management on the radio interface in order to ensure the specified quality standards (QoS) for each service provided to the majority of subscribers, in particular, for those who are in roaming.
In cellular networks of the 3rd generation, the throughput of communication channels was about a few megabits per second. To obtain high speeds, it was necessary to increase the operating bandwidth to 20 megahertz or more, which led to the emergence of 4th generation standards: LTE (Long-Term Evolution) and WiMAX (Worldwide Interoperability for Microwave Access).
The growth in the volume of multimedia mobile applications leads to the fact that the required quality of service can only be ensured by using effective methods to increase the throughput of the radio interface, since it is with wireless access that sharp load distortions occur due to stochastic movement of subscribers. The constant reduction in the cost of mobile communication services, the emergence of new types of subscriber terminals, the development of services for targeted transmission of streaming video leads to an increase in real-time traffic, the speed of which should be constant. At the same time, the Program of the Ministry of Telecom and Mass Communications "Digital Economy of the Russian Federation" predicts a sharp increase in the near future as well as traffic from machine-to-machine 7 interaction (M2M), which, as a rule, has elastic properties and allows you to change the data transfer rate within certain limits, within depending on the conditions on the network, that is, an increase in elastic traffic.
The following factors have the most significant impact on performance management in mobile networks: violation of the integrity of information of network process management commands; information blocking; violation of the logic of the software. An analysis of the results of the operation of existing LTE networks has shown that they are controlled by the automatic network management function SON, which is hardware-integrated into the equipment of the network nodes. However, this function can only introduce thresholds for the number of connections, without limiting the data transfer rate, which is not rational, in particular, when organizing roaming.
Initially, models for estimating the radio resource of mobile networks included only a homogeneous type of traffic, for which analytical solutions were found and recurrent algorithms were developed. Later, solution algorithms were developed for heterogeneous traffic having a constant rate. In connection with the emergence of applications that do not require a constant data transfer rate, elastic data traffic models have begun to be calculated. However, these studies were carried out without taking into account the access schemes that implement priority service in multiservice mobile networks of the fourth generation.
In existing and prospective mobile communication networks, the task of reducing the shortage of resources due to the emergence of services that require high transmission speed is an urgent task. Therefore, in order to increase the efficiency of the data transmission resource, it is proposed to control the rate of elastic traffic. This will allow not only to increase throughput, but also to improve the quality of service.
The LTE standard is focused only on packet traffic based on HSPA (High Speed Packet Access) technology, first implemented in UMTS (Universal Mobile Telecommunications System) networks, and EVDO (Evolution-Data Only) technology, first implemented in networksCDMA2000 standard, which are based on the idea of using transmission channels with a common frequency band, but with different pseudo-random binary sequences of the transmitter. All interfaces of the LTE network, except for the radio interface, are based on the use of the IP protocol, therefore, LTE networks are referred to as IP networks [2].
LTE is an evolution of the 3GPP UMTS standard. LTE includes the E-UTRAN (Evolved Universal Terrestrial Radio Access Network) radio access network and the new EPC (Evolved Packet Core Network) system architecture.
LTE networks are focused on using the global packet network GERAN and UTRAN for roaming [3]. The architecture of the integrated mobile communications network is shown in Figure 1.
Figure 1 - Architecture of the Integrated Mobile Network
The LTE network architecture was developed in such a way as to ensure the transmission of packet traffic with minimal delivery delays in terms of QoS [4]. In connection with the task set, it is important to increase the efficiency of packet transmission scheduling algorithms in LTE networks. In the process of ensuring QoS performance, it is necessary to perform packet scheduling to determine the order in which packets are served in a particular queue [5]. It should be noted that scheduling time plays an important role here [6].
LTE networks must support handover and roaming procedures with all existing networks, end equipment must provide ubiquitous coverage of wireless broadband services.
Burst allows multiple services to be combined [7], including voice traffic using Circuit Switched FallBack (CSFB), Over the top (OTT) and basic Voice over LTE (VoLTE). Unlike previous generation network standards, which are characterized by distributed network responsibility, the architecture of LTE networks is flat, since network interaction occurs between LTE base stations (eNodeB) and a mobility management entity (MME), usually including a network gateway – combined MME/GW devices.
A number of experiments were carried out using the simulation method. The number of generated packets N=104 / 105, the size of the packets was generated randomly. The intensity of the input stream and the intensity of service were set in the range from 0 to 1. In addition to studying the dependences of the average packet processing time, the average waiting time in the queue on various parameters, it is of particular interest to analyze the dependence of the average service time on the intensity of the input stream and the dependence of the number of served packets on the intensity input stream. 4 service strategies have been implemented: FIFO (First In, First Out) or FCFS, RR (Round Robin) [8], SJF (Shortest Job First) [9] and RED (Random early detection) [10]. The plot of the average service time versus the intensity of the input stream is shown in Figure 2, and the plot of the number of packets served versus the intensity of the input stream is shown in Figure 3. The results were obtained with a given service probability of 0.5.
Figure 2 - Graph of the dependence of the average packet service time on the intensity of the input stream
The SJF service strategy allows for the lowest average service time, this is due to the fact that short packets go ahead in the queue and do not wait for long ones to complete. The FIFO and RR queue graphs are also quite similar, the queues fill up very quickly when using FIFO and RR. Queues when using the SJF strategy fill up more smoothly. ARED allows you to keep the queues empty, so that when congestion occurs, the network device can accept additional packets and provide the required level of service, unlike other strategies. Average service time when using FIFO strategies and RR is more than 2 times higher than the identical indicator when using the SJF strategy. It should be noted that for other input values, the difference can be more than 10 times.
Figure 3 - Graph of the number of served packets versus the intensity of the input stream
The graph of the dependence of the number of served packets on the intensity of the input stream shows that all the considered strategies have denials of service, due to moderate queue filling when using the strategyand SJF this indicator is somewhat better. It should be noted that for other input values, the number of served packets may differ by more than 2 times.
Different approaches to planning, which are part of the system for ensuring the quality of service QoS in LTE networks, are analyzed. The analysis of scheduling algorithms showed that the FIFO and RR strategies have the largest average packet service time, all other things being equal, and the use of the SJF strategy allows achieving significant results compared to FIFO and RR. Queues when using the SJF strategy are filled more smoothly than when using FIFO and RR, and ARED allows you to keep the queues not filled to the end, so it is of interest to develop and use combined service strategies in LTE networks.
Literature
- Zhang X. LTE Optimization Engineering Handbook. Wiley, 2018. 844 p.
- Mousavi H. LTE physical layer: Performance analysis and evaluation // Applied Computing and Informatics, 2019. Vol. 15(1), pp. 34-44.
- Asadi A. A survey on device-to-device communication in cellular networks // IEEE Communications Surveys and Tutorials, 2014. Vol. 16(4). pp. 1801- 1819.
- Седов В.А., Седова Н.А. Самооценка системы менеджмента качества с использованием теории нечетких множеств // Программные системы и вычислительные методы, 2014. № 4. С. 456-463.
- Шапошников, Д.Е. Применение принципа гарантированного результата для учёта качественной информации о предпочтениях при комплексной оценке качества функционирования телекоммуникационных сетей // Инженерный вестник Дона, 2014, № 4. URL: ivdon.ru/ru/magazine/archive/N4y2014/2574.
- Ажмухамедов И.М., Гостюнин Ю.А. Выбор стратегии технического обслуживания и ремонта оборудования сетей связи на предприятиях нефтегазового комплекса // Инженерный вестник Дона, 2017, № 2. URL: ivdon.ru/ru/magazine/archive/N2y2017/4202.
- Сироткин А.В., Бархатов Н.И. Модель системы автоматизированного управления информационным обслуживанием // Инженерный вестник Дона, 2013, № 4. URL: ivdon.ru/ru/magazine/archive/n4y2013/2021.
- Rojas-Cessa, R., Oki E. Round-Robin Selection With Adaptable-Size Frame in a Combined Input-Crosspoint Buffered Switch // IEEE Communications Letters, 2003. Vol. 7(11). pp. 555-557.
- Peckol J. Embedded Systems: A Contemporary Design Tool. Wiley, 2019. 900 p.
- Floyd S., Jacobson V. Random early detection gateways for congestion avoidance // IEEE/ACM Transactions on Networking, 1993. Vol. 1(4). pp. 397-413