CC BY
97
Ч7
ANALYSIS OF TEMPORAL PARAMETERS OF A WIRELESS SELF-ORGANIZING NETWORK TRAFFIC SERVICE
Lev B. Buziuukov, Head of the Department of Software Engineering and Computer Science, Ph.D., Professor, Saint-Petersburg State University of Telecommunications. prof. M.A. Bonch-Bruevich, St. Petersburg, Russia, levbuz@mail.ru
Darina V. Okuneva, graduate student of software engineering and computer science, Ph.D., Professor, Saint-Petersburg State University of Telecommunications. prof. M.A. Bonch-Bruevich, St. Petersburg, Russia, darina_okuneva@mail.ru
Alexander I. Paramonov, graduate student of software engineering and computer science, Ph.D., Professor, Saint-Petersburg State Keywords: wireless self-organizing network, sensor University of Telecommunications. prof. M.A. Bonch-Bruevich, n°de, tie head node, tie quality of senke, dehy, St. Petersburg, Russia, alex-in-spb@yandex.ru loss probability, poisson fidd, gaussian f,M
One of the directions of development of the Internet of Prophetic (loT) is creation of pervasive sensor networks which basis are the self-organized wireless networks [1]. Their properties depend on a location mode of nodes, their status and the made traffic. The most important characteristic of a network is quality of functioning (service of traf-fic) which reflects a possibility of provision of different services. For the self-organized wireless networks figures of merit of service and the requirement to them are defined by purpose of a network [2]. Functioning of the self-organized wireless networks substantially is defined by their structure which depends on a method of distribution of nodes in a service zone. In this article the description of the conducted researches of temporal and structural indices of the self-organized wireless network in case of different location modes of nodes and the organization of structure of a network is this, and the received results of researches are also given. Researches were conducted by simulation modeling. During researches it is revealed that time of delivery of the message on the self-organized wireless net-work is defined by length of a route and time of delivery of a packet on sections of this route. Dependences of average length of a route are received. It is revealed that distribution functions of lengths of routes can be described by Veybull's distribution. Dependences of a median number of transit nodes on the wireless network are received. The most loaded sections of routes of a network are revealed. Dependence of length of a route on layout of the gateway in a service zone is received. Methods of the description of model of a time delay of delivery of data are given. Indi-ces necessary for the account are given in case of the organization of transit in one and different frequency channels. It is revealed that use of model of multiphase queuing system, in case of an adequate choice of its parameters, allows to evaluate with an accuracy sufficient for practical purposes time of delivery this by a route on the self-organized wireless network with rather large number of nodes. The results received in this operation can be used for further researches of functioning of the self-organized wireless networks in case of other distribution laws of nodes.
Для цитирования:
Бузюков Л.Б., Окунева Д.В., Парамонов А.И. Анализ временных параметров обслуживания трафика беспроводной самоорганизующейся сети // T-Comm: Телекоммуникации и транспорт. - 2016. - Том 10. - №10. - С. 66-75.
For citation:
Buziuukov L.B., Okuneva D.V., Paramonov A.I. Analysis of temporal parameters of a wireless self-organizing network traffic service. T-Comm. 2016. Vol. 10. No.10, рр. 66-75.
The reported study was supported by RFBR, research project No 15 07-0943la "Development of the principles of construction and methods of self-organization for Flying Ubiquitous Sensor Networks".
Introduction
Time and probabilistic traffic services are the key indicators of the operation quality of the telecommunication network. These values for public networks are standardized [3] and determined by the basic requirements provided communication services.
For self-organizing networks, these indicators are also crucial. The essential difference between these classes of networks, defining the quality of service requirements of the public communications networks is their purpose, defined by their Held of application. Requirements to quality of service traffic may vary widely, depending on ihe network task [2], For example, the requirements for probability of loss and delay may be significantly lower for networks tolerance to losses and delays (DTN) [4] than for public communication networks. These parameters depend strongly on the organization of communication between the nodes of the network and on the method of its construction, i.e., topology and routing techniques.
Methods of construction of the network and its topology are largely determined by the scope and purpose of the network. Therefore, it is advisable to be able to choose the network parameters, taking into account their impact on the time and probabilistic characteristics (TPC).
The relative position of the self-organizing wireless network nodes is crucial in terms of its potential. Using wireless communication between nodes defines the requirements for topological characteristics of the network [2].
Often traffic data service delivery network is made on some routes, which are selected by used network layer protocol that implements a particular route selection method [5J. This route may contain a number of hops, including nodes and connecting channels. The passage of each hop requires a time-consuming, which defines the data delivery delay from sender to receiver. Also, the data loss can occur in each of the hops, which probability is collectively determines the probability of data loss on the route. From these considerations it follows that the quality of service (loss and delay) depends on the route characteristics, which are the number of transits and the length of the hop. In general, these characteristics are random, therefore, to describe them, it is necessary to determine the probability distribution laws they obey.
The quality of service depends on the properties of the traffic as a process. In different purposes networks this process may have different characteristics [6], For example, the number of applications of rapidly developing technology machine-to-machine (M2M) traffic is characterized by a deterministic and a random number [7]. Video traffic characterized by the properties of self-similar process and voice traffic is close to the simplest flow [8]. These differences also have an impact on the approaches to the choice of network parameters, having a specific purpose.
Network model as a Queuing System
In considering the time and probability characteristics of the network, the subject of the analysis is traffic transmission route. The processes occurring at the physical, data link and network layers impact on TPC. Description of the influence of the physical and link layers depends on the technologies of wireless communication between network nodes. In the literature known models allowing to describe the physical and data link layer for
different wireless technologies, for example [9-11]. These articles focus on the examination of network-level models.
In general, the route can be represented as a multiphase queuing system (QS). In addition, each of the maintenance phases is the part of the route. Thus, the number of phases of the QS is the number of jumps in the route.
In general, each of the phases is a type of QS GIGIMk [13| (Fig. 1).
Fig. 1. Route model as a multi-phase QS
Message delivery lime between the source and recipient of the service depends on the parameters of the phases and their quantity. Number of service phases determined by the number of transits in the itinerary .
For the analysis of the number of transits in the route network the simulation model was constructed. It was formed by i 00 nodes located on a flat surface in the area bounded by the square of 200x200 m. The radius of the communication node is R = 50 m.
The length of the random route (point to point)
In the investigation a random distribution of nodes on the territory was examined: for uniform distribution law independent x and y coordinates and normal distribution (Fig. 2).
Using the normal distribution of the dispersion value is chosen to be a dispersion of uniform distribution [14]
iT'
12
(1)
where b, a — the boundaries of the random variable (in this example a = 0, b = 200 m )•
To find the shortest paths between all pairs of nodes F!oyd algorithm was used [15]. Figure 3 shows the empirical histogram of the lengths of the shortest routes (including transits) and approximation of probability density of the Weibull distribution [14].
(2)
where a and 0 - the parameters of the distribution.
Obtained form distribution of the number of transits and their average values (6,5 ±0,5), shown on fig 3, statistically equal for
uniform and normal distribution nodes coordinates. The simulation results showed, that the average number of transits (jumps) in the route depends on the dispersion.
. J • *»
; ... •
• .. и ■ * * * 1
• л
ш »
f
Fig. 4 shows the average number of transits (jumps) in the path of the ratio of the standard deviation to the radius of the
communication node .£., where a = -Jd, D — dispersion, R
R - radius of the communication node.
0/
> v = 11,4761
0,2
0,4
0,6 c/R
1,2
Fig. 4. Average number of transits in the route of the standard deviation
The increase in the dispersion leads to increase in the average number of transits by law close to linearly. The experiment results obtained empirical dependence of the average number of transits of the dispersion:
m =
J R , else
(3)
Fig. 2. Possible distribution units on the territory: the uniform and normal distribution of the coordinates (for equal variances)
A
I
4 Б S ID 12 M 1Б la 22 24 26 28 82 34 Число транзитов
0 1 Л Б 8 10 II 1J 16 1! 20 24 28 30 32 34 Число транзитов
Fig. 3. The number of transits in the itinerary for the uniform and normal distribution laws (with equal variances)
where £ = 4^43 (according to the simulation).
The simulation results showed that the average number of transits increases until the disconnected state moment of phase transition in the network [16J.
These results allow us to estimate the average number of jumps in a random route, i.e., the route between any two nodes of the network. In a number of tasks required to make delivery of the traffic not only between two random nodes (point-to-point), but between the target node (or group of units) and the rest of the network nodes (point-to-multipoint or muItipoint-to-point). These options are discussed further routes.
Route in a network gateway (multipoint-to-point)
Often, the network is built in applications for data collection, obtained from a plurality of points distributed in some way on the surface or in space. Network task is sending data from the plurality of points (nodes) to the gateway, which performs the function of data collection or to another transit network, or the processing means. Typically, the network gateway is located in the geometric center of the coverage area. Then the data direction oriented radially from the edge of coverage to its geometric center zone (multipoint-to-point). Using the simulation model, described above, the distribution route lengths were obtained from the network nodes to the gateway located at the geometric center of the coverage area. The results are shown in fig. 5.
As seen from the drawings, the average transit center to the gateway is smaller than in the case of point-to-point (Fig. 4), and is equal to a statistically normal distribution and uniform node coordinates in the coverage area (3,8±0,5 nodes) ■
m
MV
H-7
MV
and intensity of the load on the various phases of the service time and the properties of the traffic flow.
As shown above, the average number of maintenance phase (the number of the route transits) dispersion depends in the most degree on the coordinates of nodes and can be calculated according to the empirical relationship (3), with the result rounded to the higher whole value.
The properties of traffic flow, arriving at the maintenance phase, depend on the functions to be solved by the network. In the most cases, order to monitor, used the periodic data transmission with a ccrtain period t - In this case, traffic produced by one node is deterministic, but for the asynchronous network nodes, ilie aggregated traffic flow has random characteristics.
The intensity is determined by the payload traffic intensity of one sensor node i and the average number of directions dri,
served by one transit hub. Thus, for the i- th phase service
A A = K' K + ^i-i' t-2...m, (5)
where k a 2,5 - according to the results obtained in the simulation.
However, along with the nodes served by the i - th transit point, in the communication zone or zone of mutual influence (interference) may be other transmitting nodes. If multiple nodes are at a distance from each other less than the radius of the mutual influence (interference) when they are working on one frequency channel, the communication channel capacity is reduced. This occurs because of the need to distribute the using of a common time-frequency resource of the interference signals. Sometimes for solution of this problem are used protocols that enable you to receive and transmit data on different frequency channels [20J. In this case, the transmission of the packet spends lime about tw ice more and the bandwidth is reduced about by half [19].
In the case of using one frequency channel bandwidth is reduced by a factor equal to the number of nodes [ 19] in the interference zone. Assume that reporting transit node idle during affecting activity units time, then, in context of the model, this situation is equivalent to the situation when traffic intensity at the input node is equivalent to the total traffic intensity produced by nodes in the interference zone A,+JL,in ■ This value for the case of
uniform distribution of nodes can he estimated (Poisson liekls) as:
X.=A = !TR]p-,i,,, (6)
where R - radius of the transit point interference zone (in general, it may differ from the communication node radius R)\ p -density of nodes in the interference region (the number of nodes located on the unit area); ^ - intensity of the traffic produced by
one unit (the number of packets per unit time.).
In case of normal distribution nodes (Gaussiau field) the number of nodes in the interference zone is determined
= pj -Aq, (7)
where n - total number of nodes; pj - likelihood of the node
entering to the interference zone, which for a Gaussian field is generally defined as the integral of the density distribution function of coordinates on the border of the zone of interference }j/(v y)dxdy' Pract'ca' purposes, in some cases, it may be
I
assumed constant density of nodes within the boundaries of the interference zone [21 ].
In the case of using different frequency channels, the effect of interference is eliminated, and the intensity of the load on the Hub entrance will be determined in accordance with (5).
Average service time for i - th phase is determined by transmission speed of a communication channel b„ bit/s, and by the
way of communication organizing. Execpt the time to transmit a packet, mechanism to prevent the of collision MAC level type involves the introduction of protective interval, random length (backoff - interval). It eventually leads to increase in transmission time to an average value of the interval, i.e., its duration depends on the probability of collisions, as well as the time required to transmit an acknowledgment.
Along with the time needed for performing operations employed transmission protocol, different modes of network nodes may be used in order to save energy. This node can accept a message passed just over a periodically repealing interval (active node status interval). This leads to waiting for the active stale and increases the transmission time [23].
t=tMAC + f,
where i\/¡^ - time required to perform used transmission protocol operations (MAC level); f - average waiting time for the active node status.
For example, for the standard IEEE 802.14.5 [25] the average transmission time for free channel, according to [25], can be defined as
hi AC ~ Tbackoff + Tdata + Tack + Ttrt where T^^jr ~ backoff interval time (default 2,368 ms); T^ata ~
data transfer time (4.256 ms); j . - transmitting an acknowledgment message time (0.352 ms); yj ~ standby time of
confirmation (0.192 ms); t - waiting time of active node status.
If the transit node is used for one frequency channel, the maintenance time is determined by the data rale in the channel ¡19] and the additional time required to perform the operations of defined transmission protocol
t fmac ~ t MA C' ^
If different transmit and receive frequency channels are used by the transmission node [20], the bandwidth is about half of the potentially achievable data rale over the channel. If we consider the last option, the packet delay on one phase of service will be defined as
'mfmac = 2/~A1AC? (4')
The average waiting time of the active stale of the node specifically for a specific application and can vary widely.
To estimate the waiting time at the i - th phase of the service we w ill use the approximate formula ofG/G/l system [13]
<7j] + t7," t2
(10)
where yi = Zj • t - the intensity of the load (Earl); t - average service time (s); ÏÉ__L — time interval between the messages
(packets) (s); aB - standard deviation time interval between messages (s); eTt - standard deviation of the service time (s).
My
spread values, which is caused by a sharp increase in the number of unsuccessful (retransmission) packets.
O - - r. T-| -r IÛ r - 'X
Fig. 13. implementation of the traffic at the gateway (the number of packets per interval I s)
0,20 o,ia
0,15 0,14 0,12 0,10 o,os 0,06 0,04 0,02 0,00
I
I I
1 normal disti ilMrtion
:x * It
pci^1
uniform distribution
Fig. 14. Distribution of the number of incoming packets on a network gateway
This result, based on the properties of the elementary stream, allows selecting the magnitude of the standard deviation of time intervals between packets cr/; in the model (10), which in this
case is an average interval. It should be noted that the flow properties are dependent on the nature of the traffic produced by network nodes and generally can vary from a simple stream properties. Therefore, the result should be considered only in the context of this model.
Fig. 15 shows a plot of the delivery time from the sensor package to the gateway node. The graph shows the theoretical (10) curves and simulation results for two types of distribution of coordinates on the territory of the nodes: the uniform and normal. In both cases the same node values selected coordinate dispersion. As seen from the increased traffic intensity dependency increases the delay of delivery of data. Comparative analysis of the results for different types of distribution coordinates of the nodes shown that in the case of normal distribution, with relatively high values of load, there is a large delay in packet delivery. This is due to a greater load on the last leg in the route appears at the gateway to the greatest extent. More in comparison w ith uniform distribution of the load intensity is due to coordinate the uneven density of nodes in the case of a normal distribution of the coordinate that has the greatest value in the center of the coverage area, i.e., placing at the gateway. At high load (overload), the simulation results showed a significant delay
Fig. 15. Delivery delays depending on the traffic intensity for uniform and normal distributions of the nodes coordinates (the average length of the number of transits 3)
In the context of small and medium load value delivery delays are statistically similar for both types of distributions coordinates.
Comparison of the results of simulation and analytical modeling indicated that they are close enough (deviation is no more than 20%) in the field of small and medium loads. Significant deviations are observed in the area of the load values at which the gateway overload occurs.
Thus, the use of a multiphase model QS, with adequate choice of its parameters, allowing a sufficient accuracy for practical purposes to evaluate time data delivery route network with a sufficiently large number of nodes.
Conclusion
As a result of studies described in this article we can do the following conclusions:
• Time of delivering a message to the FSU determined by the length of the route and time of arrival on the sections of the route;
• The distribution function of routes lengths can be described by Weibull distribution, as for random routes in the network (point-to-point) and how so for the destinations (multipoint-to-point) to the gateway, located in the center of the serv ice area;
• The average length of the route depends on route node coordinates and is independent of the distribution of nodes in the territory. The increase in the dispersion leads to the growth of the length of the route, until the decline of network connectivity. This relationship with sufficient for practical calculations accuracy described by a linear function;
• If the gateway locates in the center of the service area than the average path length is less than to random route;
• The average number of transit nodes on the network depends on the dispersion node coordinates and is independent of the type of distribution service area. With the growth of the dispersion share of the transit nodes increases up to the moment of reducing network connectivity. This relationship with sufficient for practical calculations accuracy described by a linear function;
• The number of destinations served by transit nodes is unequally. The most loaded section of the route is the last section of the route between the node and the network gateway. The least loaded are the initial parts of the routes;
• Delayed delivery model of data sources to the network gateway may be described as a multiphase SMO, wherein each of the maintenance phase must take into account reduction in throughput due to interference signals and operation modes of network nodes;
• When the transit organization on a single frequency channel, to account for the interference signals in the model of each of the maintenance phase should take into account the traffic affecting network nodes within the area of the transit node of interference;
• When the transit organization on different frequency channels should take into account the increase in packet service time due to lower bandwidth channels transit nodes;
• queuing system model as a model for each of the maintenance phase may be used upper limit of the analytical model timeout G/G/l.
References
1. Kucheryavy A.B. Internet of Things / Electrosvyaz, 2013. No.i.
2. Kucheryavy A.E.. Prokopev A. V.. Kucheryavy E.A. Self-organizing networks. St. Petersburg, 201 1, 316 p.
3. ITU-T Network performance objectives for IP-based services. Recommendation Y, 1541.
4. RFC 4838 De I ay-Tolerant Networking Architecture. April 2007.
5. Neha Rathi. Jyoti Saraswat, Partha Pratim Bhattacharya. International Journal of Distributed and Parallel Systems (IJDPS). Vol. 3, No.5, September 2012.
6. Shelukhin O.I., Osin A. V.. Smolsku S.M. Self-similarity and fractals. Telecommunication applications. Moscow: FIZMATL1T. 2008. 368 p.
7. Paramonov A.!. Models of traffic flows for M2M networks / Electrosvya2,2014. No. 4. Pp. 9-14.
8. Paramonov A.I. Development and research of complex traffic models for public communication networks. Abstract. 2014.
9. Giuseppe Bianchi. IEEE 802.11-Saturation Throughput Analysis. IEEE Communications Letters. Vol. 2. No. 12, December 1998.
10. G. Bianchi. Performance Analysis of the IEEE 802.11 Distributed Coordination Function. IEEE Journal on Selected Areas in Communications, 2000. Vol. 18. No. 3.
11. Timo Vanhatupa. Wi-Fi Capacity Analysis for 802.llac and 802.1 In: Theory & Practice. Ekaliau Wi-Fi Design White Paper, 2013.
12. Marina Petrova. Study of IEEE 802.15.4 Using Measurements and Simulations, Wireless Communications and Networking Conference, 2006. WCNC 2006. IEEE (Volume: 1).
13. Kleinrock L. Queuing theory. Moscow. Mechanical engineering, 1979. 432 p.
14. WentzelE.S. Probability. Moscow. Nauka, 1969. 576 p.
15. Christofides N. Graph Theory. An algorithmic approach. Moscow. Mir. 1978.430 p.
16 Raigorodskii A.M. The models of random graphs and their applications. Proceedings of MIPT. 2010. Vol. 2. No 4.
)7. Kirichek R.V.. Paramonov A.I.. Prokopiev A.V.. Kucheryavy A.E. The evolution in the field of wireless sensor networks research. Information technology and telecommunications. 2014. No. 4(8). Pp. 29-4!.
18. Muthanna A., Paramonov A., Koucheryavy A., Prokopiev A. Comparison of Protocols for Ubiquitous Wireless Sensor Network, International Congress on Ultra Modern Telecommunications and Control Systems and Workshops 6. Ser. "2014 6th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops, 1CUMT20I4", 2015. Pp. 334-337.
19. Osteriind Fredrik, Ditnkels Adam. Approaching the maximum 802.15.4 multi-hop throughput. Swedish Institute of Computer Science Publications Database, 2008.
20. G. Zhou, C. Huang. T. Yan. T. He, J. Stankovic, and T. Abdelzaher. Mmsn: Multi-frequency media access control for wireless sensor networks. In INFOCOM, 2006.
21. Buziuukov L.B., Okitneva D.V., Paramonov A.I. Connectivity analysis of self-organizing w ireless network at various distribution sites around the area, Electrosvya2.2016. No, 9.
22. A. Dunkels, B. Gronvall. T. Voigt. Contiki - a lightweight and flexible operating system for tiny networked sensors. 29th Annual IEEE International Conference on Local Computer Networks, 2004.
23. Adam Dunkels. The Contiki MAC Radio Duly Cycling Protocol. S1CS Technical Report T2011: 13 ISSN 1100-3154 December 2011.
24. Joakim Eriksson. Detailed Simulation of Heterogeneous Wireless Sensor Networks. Dissertation lor the degree of Licentiate of Philosophy in Computer Science at Uppsala University in 2009,
25. IEEE 802.15.4 Specification (2006) [SS95552],
26. Calculating 802.15.4 Data Rates.NXP Laboratories UK Ltd, JN-AN-1035 (vl. I ) S-Nov-2013.
27. RFC 6550. IPv6 Routing Protocol for Low-Power and Lossy Networks. Internet Engineering Task Force (IETF), March 2012.
28. Muthanna A.. Paramonov A.. Koucheryavy A.. Prokopiev A. Comparison of Protocols For Ubiquitous Wireless Sensor Network. 6th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops, 1CUMT20I4. 2015. Pp. 334-337.
Ч7
АНАЛИЗ ВРЕМЕННЫХ ПАРАМЕТРОВ ОБСЛУЖИВАНИЯ ТРАФИКА БЕСПРОВОДНОЙ САМООРГАНИЗУЮЩЕЙСЯ СЕТИ
Бузюков Лев Борисович, профессор, зав. кафедрой программной инженерии и вы-числительной техники, Санкт-Петербургский государственный университет телекоммуникаций им. проф. М. А.Бонч-Бруевича (СПбГУТ),
Санкт-Петербург, Россия, levbuz@mail.ru Окунева Дарина Владимировна, аспирант, Санкт-Петербургский государственный университет телекоммуникаций им. проф. М. А.Бонч-Бруевича (СПбГУТ), Санкт-Петербург, Россия, darina_okuneva@mail.ru Парамонов Александр Иванович, Профессор кафедры сетей связи и передачи данных, Санкт-Петербургский государственный университет телекоммуникаций им. проф. М. А.Бонч-Бруевича (СПбГУТ), Санкт-Петербург, Россия,
alex-in-spb@yandex.ru
Аннотация
Одним из направлений развития Интернета Вещей (IoT) является построение всепроникающих сенсорных сетей, основой которых являются самоорганизующиеся беспроводные сети [1]. Их свойства зависят от способа размещения узлов, их состояния и производимого трафика. Важнейшей характеристикой сети является качество функционирования (обслуживания трафика), которое отражает возможность предоставления различных услуг. Для самоорганизующихся беспроводных сетей показатели качества обслуживания и требования к ним определяются целевым назначением сети [2]. Функционирование самоорганизующихся беспроводных сетей в значительной степени определяется их структурой, которая зависит от способа распределения узлов в зоне обслуживания. Дано описание проведенных исследований временных и структурных показателей самоорганизующейся беспроводной сети при различных способах размещения узлов и организации структуры сети, а так же приводятся полученные результаты исследований. Исследования проводились путем имитационного моделирования. В ходе исследований выявлено, что время доставки сообщения в самоорганизующейся беспроводной сети определяется длиной маршрута и временем доставки пакета на участках этого маршрута. Получены зависимости средней длины маршрута. Выявлено, что функции распределения длин маршрутов могут быть описаны распределением Вейбулла. Получены зависимости среднего числа транзитных узлов в беспроводной сети. Выявлены наиболее загруженные участки маршрутов сети. Получена зависимость длины маршрута от расположения шлюза в зоне обслуживания. Приведены способы описания модели задержки доставки данных. Приведены показатели необходимые для учета при организации транзита на одном и разных частотных каналах. Выявлено, что использование модели многофазной системы массового обслуживания, при адекватном выборе ее параметров, позволяет с достаточной для практических целей точностью оценить время доставки данных маршрутом в самоорганизующейся беспроводной сети с достаточно большим числом узлов. Полученные результаты могут быть использованы для дальнейших исследований функционирования самоорганизующихся беспроводных сетей при других законах распределения узлов.
Ключевые слова: беспроводные самоорганизующиеся сети, узел датчика, качество обслуживания, пуассоновское поле, гауссовское поле. Литература
1. Кучерявый A.E. Интернет вещей // Электросвязь, 2013. №1.
2. Кучерявый A.E., Прокопьев А.В., Кучерявый Е.А. Самоорганизующиеся сети. Санкт-Петербург, 2011, 316 с.
3. ITU-T Network performance objectives for IP-based services. Recommendation Y.1541.
4. RFC 4838 Delay-Tolerant Networking Architecture. April 2007.
5. Neha Rathi, Jyoti Saraswat, Partha Pratim Bhattacharya. International Journal of Distributed and Parallel Systems (IJDPS) Vol. 3, No.5, September 2012.
6. Шелухин О.И. Осин А.В., Смольский С.М. Самоподобие и фракталы. Телекоммуникационные приложения. - М.: Физматлит, 2008. - 368 p.
7. Парамонов А.И. Модели трафика для M2M сетей // Электросвязь, 2014. № 4. С. 9-14.
8. Парамонов А.И. Разработка и исследование сложных моделей трафика для сетей связи общего пользования. Автореферат дисс. доктора техн. наук, 2014.
9. Giuseppe Bianchi. IEEE 802.11-Saturation Throughput Analysis. IEEE Communications Letters, VOL. 2, No. 12, December 1998.
10. G. Bianchi. Performance Analysis of the IEEE 802.11 Distributed Coordination Function. IEEE Journal on Selected Areas in Communications, 2000, vol. 18, № 3.
11. Timo Vanhatupa. Wi-Fi Capacity Analysis for 802.llac and 802.lln: Theory & Practice. Ekahau Wi-Fi Design White Paper, 2013.
12. Marina Petrova, Study of IEEE 802.15.4 Using Measurements and Simulations. Wireless Communications and Networking Conference, 2006. WCNC 2006. IEEE (Vol. 1).
13. Клейнрок Л. Теория массового обслуживания. - M.: Машиностроение, 1979. - 432 с.
14. Wentzel E.S. Теория вероятностей. - M.: Наука, 1969. - 576 с.
15. Christofides N. Теория графов. Алгоритмический подход. - M.: Mир, 1978. - 430 с.
16. Райгородский А.М. Модели случайных графов и их приложения. Труды МФТИ, 2010. Том 2. № 4.
17. Киричек Р.В., Парамонов А.И., Прокопьев А.В., Кучерявый А.Е. Эволюция в области исследования беспроводных сенсорных сетей // Информационные технологии и телекоммуникации, 2014. № 4 (8). С. 29-41.
18. Muthanna A., Paramonov A., Koucheryavy A., Prokopiev A. Comparison of Proto-cols for Ubiquitous Wireless Sensor Network. International Congress on Ultra Modern Telecommunications and Control Systems and Workshops 6. Ser. "2014 6th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops, ICUMT 2014" 2015. pp. 334-337.
19. Оsterlind Fredrik, Dunkels Adam. Approaching the maximum 802.15.4 multi-hop throughput. Swedish Institute of Computer Science Publications Database, 2008.
20. G. Zhou, C. Huang, T. Yan, T. He, J. Stankovic, and T. Abdelzaher. Mmsn: Multi-frequency media access control for wireless sensor networks. In INFOCOM, 2006.
21. Бузиюков Л.В., Окунева Д.В., Парамонов А.И. Анализ связности самооргани-зующейся беспроводной сети при различном распределения узлов по территории // Электросвязь, 2016. №9.
22. A. Dunkels, B. Gronvall, T. Voigt. Contiki - a lightweight and flexible operating sys-tem for tiny networked sensors. 29th Annual IEEE International Conference on Local Computer Networks, 2004.
23. Adam Dunkels. The Contiki MAC Radio Duty Cycling Protocol. SICS Technical Re-port T20ll: 13 ISSN 1100-3154 December 2011.
24.Joakim Eriksson. Detailed Simulation of Heterogeneous Wireless Sensor Networks. Dissertation for the degree of Licentiate of Philosophy in Computer Science at Uppsala University in 2009.
25. IEEE 802.l5.4 Specification (2006) [SS95552].
26. Calculating 802.l5.4 Data Rates. NXP Laboratories UK Ltd, JN-AN-l035 (vl.l) 5-Nov-20l3.
27. RFC 6550. IPv6 Routing Protocol for Low-Power and Lossy Networks. Internet Engineering Task Force (IETF), March 20l2.
28. Muthanna A., Paramonov A., Koucheryavy A., Prokopiev A. Comparison of Protocols For Ubiquitous Wireless Sensor Network. 6th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops, ICUMT 20l4, 20l5. pp. 334-337.
