549_Sovremennye_problemy_telekommunikatsij_
.pdfЛитература
1.1. «A Novel SLA Framework for Time-Differentiated Resilience in Optical Mesh Networks» M. Xia, M. Tornatore, S. Sevilla et al. IEEE Journal of Optical Communications and Networking, Vol. 3, Issue 4, 2011.
Колягин Леонид Валерьевич
аспирант по направлению "Электроника, радиотехника и системы связи" СибГУТИ
(630102, Новосибирск, ул. Кирова, 86) тел. (8-383) 203-4945 (доб. 4090), e-mail: LKolyagin@nvg.ru.
Algorithms of connection assignment for time-differentiated resilience
L.V. Kolyagin
In this paper algorithms for connection assignment for time-differentiated resilience are submitted.
Key words: reliability, availability, critical window.
31
An Influence Study of Repeated Call Attempts on the Performance of Multirate Loss Networks with State-Dependent Routing
V.I. Meikshan, V.P. Korchagin
A non-hierarchical telecommunication network is studied with Least Loaded Routing (LLR) method related to traffic distribution. Multiple traffic flows with different bandwidth requirements are taken into account. The connection level performance metric under consideration is the end-to-end call blocking probability, especially with the influence of repeated attempts to get required service. An approximate analytical model is presented and the accuracy of this model is tested by discrete event simulation.
Keywords: multi-service network; adaptive routing; repeated calls; blocking probability.
1. Introduction
This report is devoted to the non-hierarchical multi-service network (MSN), built on the basis of packet switching technologies. Such networks are characterized by extensive application of statedependent (adaptive) routing. An important type of corresponding methods is the Least Loaded Routing (LLR), when at the stage of virtual connection setup the route with minimal load is always selected. According to this approach, among the all available routes the path will be chosen with the largest amount of free channel resources between final points. Operation experience of existing telecommunication networks argues that such routing concept prevents congestion of shortest paths, allows overloaded parts of the network to be avoided and helps traffic flows to be distributed more uniformly [1].
Repeated calls may be mentioned as a quite important factor which affects the quality of services and network performance. Under high persistence, the subscriber frequently makes repeated attempts to access the service that he wants, as long as he gets a denial of required service. Particularly, some new developments in telecommunication technologies (for example, the wide use of au- to-repeat facilities) lead to a substantial strengthening of the retrial phenomenon, which may significantly increase the tra c and degrade the network performance, especially under overload conditions. As a consequence, the quality of services may be dramatically reduced.
In many known queuing models retrials phenomenon is not taken into account. As a consequence, follow-up methods for network performance analysis involve overly optimistic grade of service estimation for incoming calls, and leads to significant mistakes at the stage of network dimensioning.
These negative circumstances require further complexity of mathematical models to more adequately describe the process of interaction between telecommunication network and subscribers. Presented examination is based on a model of similar type which allows approximate evaluating the basic characteristic of the network performance at the level of virtual connection setup – the blocking probability for incoming call (i.e. the probability of call to be abandoned due to the inability to provide a transmission path with required bandwidth). The adequacy and accuracy of constructed analytical model are validated by discrete event simulation.
32
2. A formal description of the study object and basic approach to analysis
The MSN under consideration has a mesh topology with a set of routers (nodes for packet switching), connected by a set of L digital lines (DLs). It is assumed that these lines are numbered in
optional sequence, and DL with number l (l 1,L) has a capacity Cl (in basic bandwidth units [2]) to
ensure a transmission of digital data flow with corresponding maximum speed.
The network supports different classes of traffic. Particularly, incoming traffic flow with number k (TFk; k 1,K ) relates to some pair of network end-points and the following assumptions are made: 1) primary calls (requests) on service delivery arrive with constant intensity k ; 2) virtual
connection to the destination point must have capacity bk (in basic bandwidth units) to maintain required data transmission speed; 3) a single telecommunication session (and holding time for all network resources dedicated to accepted call) has mean duration hk.
All allowable routes of packets delivery for the TFk are combined into the set Mk { km}, where a single route (path) km is described by the numbers sequence of lines which are the components of this route. In case of non-hierarchical network with mesh topology the set Mk includes
the shortest (or the direct) route between considered network end-points and also alternative routes with transit through one node.
It is assumed that selection of individual route km Mk for incoming calls which are related to the TFk, takes place with probability qkm, and
km Mk qkm 1.
With probability km selected route km has no free channel resources which are necessary for re-
quired service provision, and therefore the arriving call must be refused (blocking). Then for TFrk the probability of incoming calls blocking can be calculated as follows:
Bk km Mk qkm km. |
(1) |
Formula (1) is a common approach for the networks with a fixed (static) multipath routing, when the probabilities {qkm} are predetermined and actually express the proportionality coefficients for the flow TFk distribution within the set Mk of allowable routes. The calculation of these coefficients usually occurs on various metrics, the values of which are assigned to individual routes, and as one of such metrics the average load of the route is often used. The next chapter is devoted to a more complex procedure where considered coefficients are calculated on the basis of probability distributions, which more fully describe the stochastic nature of the workload related to individual network links. In the case of stationary (steady) operation mode of the network under consideration this formula allows to take into account some features inherent to the algorithm of adaptive (dynamic) routing.
3. Fixed-point approximation
In addition to qkm, presented in [3] mathematical model for MSN with state-dependent routing centers around the following set of basic variables: aik – the probability that the i-th digital line
(l 1,L) has a free channel resources which size is not less than bk bandwidth units; jk – total in-
tensity of incoming k-class calls for the j-th DL (this combined traffic flow is reduced by blockings on other links); pj (n) – the stationary occupancy probability of j-th DL, i.e. the probability that ex-
actly n bandwidth units are being used on this DL. Under the assumption that calls of different classes arrive at j-th DL as a Poisson processes with corresponding intensities, the link occupancy prob-
33
abilities { pj (n);n 0,Cj } may be calculated on the basis of multidimensional Erlang distribution for multi-rate multi-class multi-resource loss models [2].
Fig. 1
Interrelations between these variables are graphically expressed in fig. 1 and also may be written as a system of nonlinear algebraic equations [3] in which unknowns are jk , a jk , pj (n) and qrm .
Numerical decision of this system (i.e. the equilibrium fixed point) can be found by the method of repeated substitutions. When a solution is obtained, it is offered at first to calculate
km 1 ajk j km
and then finally by the formula (1) we can get Bk – the blocking probability of calls which relate to TFk.
4. Approximation for loss network with retrials
To characterize the behaviour of the subscriber when he persistently attempts to gain access to the required services, we will use the following parameters which quite properly take into account the most important factors related to retrials phenomenon [4]: 1) retrial intensity for one traffic source ( rep); 2) probability of repetition after the very first unsuccessful attempt (H1) and for any next repetition (H2).
Analysis of nowadays calculation methods for systems with repeated calls shows that simplified assumptions have widespread use when the aggregated flow of primary (fresh) and repeated calls is considered as the Poisson flow. For simple queuing systems it is theoretically proved that such an approximation is asymptotically exact in the case when rep 0. Applicability of this approach for network calculations is argued in [5].
To approximately replace non-Poisson flow of repeated calls by a Poisson flow it is very convenient to use equations, called as “queuing conservation laws” [6]. When be written for any system with repeated calls, such formulas have a simple physical meaning and relate together the intensities of input and output flows. In particular, under the condition of statistical equilibrium, the intensity of incoming repeated calls from Nk sources creating an input flow TFk equals to the intensity of events associated with creation of these sources [7]:
Nk rep k Bk H1 Nk rep Bk H2.
Blocking probabilities Bk and Bk in this equation are related to primary and repeated calls re-
spectively. When rep 0, the properties of repeated calls flow just slightly differ from ones for a
Poisson flow with some unknown intensity . Therefore, these GoS parameters may be approxi- k
mated by relevant probabilistic characteristics of the simplified model where such characteristics
34
have a meaning of total call losses (for a common flow of primary and repeated calls):
Bk Bk Bk .
With no doubt the error of such approximation would be lower if for estimations to be found an expression may be written which looks like a given above “conservation law”:
k k Bk H1 k Bk H2.
Hence it is easy to obtain an equation |
|
|
|
|
|
|
Bk H1 |
|
|
|
|
k k 1 |
|
|
|
, |
(2) |
1 B H |
|
||||
|
k |
2 |
|
|
|
where k k k – the total arrival rate of primary and repeated calls which relate to the flow TFk. Blocking probabilities (Bk ; k 1,K ) for incoming calls depend on the total traffic imposed upon the network and determined by the vector 
k ;k 1,K 
, so that equations (2) for all traffic flows {TFk; k 1,K } jointly compose a system of implicit equations with unknown variables k (k 1,K ). Solution of this system may be found, for example, by iterative method. At the beginning of calculations it is assumed that 
(0)k 
k 
and current s-th iteration (s 1) includes the following
actions:
1. By means of calculation procedure briefly sketched in section 3 and fully described in [3] probabilities (Bk(s) ;k 1,K ) are computed for traffic flows specified by the vector 
(ks 1) 
.
2. Using (2) elements of the vector 
(ks) 
are obtained.
A terminating condition for the described iterative process is inequality * , where * is permissible absolute error of the solution and maxk (ks) (ks 1) . Final values (Bk( ) ;k 1,K )
obtained in the last of the executed iteration give an approximate evaluation of performance characteristics for a multiservice network with state-dependent routing in the presence of repeated call attempts.
5. Experimental results
The proposed analytical model is approximate one, thus the results of the analytical calculations need to be validated by simulation experiments. Such experiments were executed with the help of software which implements discrete event simulation (DES) in accordance with the concept of object modeling [6].
The topology was chosen of a fully connected MSN with 5 nodes (Fig. 1). As can be seen from this figure, the network consists of 10 digital lines (they all have equivalent capacity C=100). It should be noted that the same network previously was investigated in [3], but without the effect of repeated calls.
Each traffic flow TFk (k 1,K) refers to a pair of network terminal points and is described by the following numerical characteristics: arrival rate of primary calls ( k ); capacity requirements
(bk), measured in the number of Basic Bandwidth Units (BBU). From the viewpoint of overall traffic volume the following regimes for network operation are considered: nominal traffic (NT); 1.2 times the NT; 1.4 times the NT; 1.8 times the NT and 2 times the NT.
In the case of NT for all considered flows (K=27) values k belong to the range 0.1 – 25. Furthermore, these flows are separated into three classes according to the type of services: low-speed
35
services (bk =1); middle-speed services (bk =2); high-speed services (bk =3). One direct route and three two-tier routes (with one transit node) are available for each traffic flow.
As a set of parameters describing the user behavior when an attempt to get the service has failed, we will use repetition rate ( rep) and two probabilities of repetition: after the first attempt (H1) and after any repeated attempt (H2). In this experiment we set H1=H2=0.999.
Fig. 1. Network topology
In the table 1 some numerical results are given which enable to judge on the accuracy of the approximation proposed. The average call blocking probability for the network is considered as a main characteristic to observe:
|
|
|
K |
|
B |
k bk Bk |
. |
||
K |
||||
|
|
|
k 1 |
|
|
|
|
k bk |
|
|
|
|
k 1 |
|
Table 1. Call blocking probability analysis (Network operation regime)
Network operation |
Call blocking probability |
Relative |
|
regime |
Analytical model |
DES |
deviation |
1.2 times the NT |
0.1423 |
0.1373 |
3.68% |
1.4 times the NT 2 |
0.4069 |
0.386 |
5.42% |
1.8 times the NT |
0.8086 |
0.7994 |
1.14% |
2 times the NT |
0.9118 |
0.9383 |
2.82% |
From the table 1 it can be seen that in various network operation regimes analytical model ensures quite accurate estimates which are sufficiently close to exact blocking probabilities obtained by DES. Thus, relative difference does not exceed 6%.
6. Conclusion
This report presents an analytical model which allows blocking probability determination for telecommunications network with adaptive routing and multiple traffic classes. The proposed calculation procedure makes it possible to analyze the influence of repeated call attempts on the network performance and the quality of services offered. The adequacy of the devised approach was determined by comparing the numerical solution of the performance evaluation algorithm with simulation results. An interesting question for future work would be to use capabilities of the model under investigation in providing performance bounds and for solving design tasks.
36
References
1.Medhi D., Ramasamy K. Network Routing: Algorithms, Protocols, and Architectures. Morgan Kaufmann Publishers, 2007. 832 p.
2.Ross K.W. Multiservice Loss Models for Broadband Telecommunication Networks. SpringerVerlag: London, Berlin, New-York, 1995. 343 p.
3.Liu M., Baras J.S. Fixed Point Approximation for Multirate Multihop Loss Networks with State-Dependant Routing // IEEE/ACM Trans. on Networking. 2004. V. 12, № 2. P. 361–374.
4.Falin G.I., Templeton J.G.C. Retrial queues. Springer, 1997. 320 p.
5.Takahara G.K. Fixed point approximations for retrial networks // Probability in the Engineering and Informational Sciences. 1996. V. 10, № 2. P. 243-259.
6.Ryzhikov Y. Laws of conservation in the queuing theory // Multiple Access Communications. Springer: Berlin, Heidelberg, 2010. P. 119-126.
7.Stepanov S.N. Generalized model with repeated calls in case of extreme load // Queuing systems. 1997. V. 27, № 1-2. P. 131-151.
37
Системы ВЧ связи по ЛЭП. ВЧ связь в сетях высокого напряжения (35-750 кВ)
И.И. Павлов
В работе будет рассмотрена высокочастотная связь по линиям электропередач высокого напряжения (35 – 750 кВ). На примере цифрового оборудования «PowerLink» компании «Siemens»будет показана схема организации ВЧ связи и особенности этой системы.
Ключевые слова: ВЧ связь, ЛЭП, «PowerLink», напряжение, скорость, передача информации.
Во время бурного развития информационных технологий (90-е гг.) предприятия электроснабжения в промышленно развитых странах делали значительные инвестиции в прокладку линий оптической связи (ВОЛС) по воздушным линиям высокого напряжения в надежде обеспечить себе прибыльную долю перегретого рынка телекоммуникаций. В это время добрую старую технологию ВЧ похоронили заново. Затем раздутый информационно- технический пузырь лопнул, и во многих регионах наступило протрезвление. И именно в энергетических сетях установка оптических линий была приостановлена по экономическим соображениям, а технология ВЧ связи по воздушным линиям приобрела новое значение.
В результате применения цифровых технологий на высоковольтных сетях, сформировались новые требования к ВЧ системам.
На ответвлениях сети и длинных участках линий электропередач использование ВОЛС экономически не целесообразно. Здесь технология ВЧ предлагает экономичную альтернативу для передачи речи, данных и сигналов-команд РЗ и ПА (РЗ — релейные защиты, ПА — противоаварийная автоматика).
Рис. 1
В связи c быстрым развитием систем автоматизации электроэнергетики и цифровых широкополосных сетей на магистральных линиях, изменились требования к современным системам ВЧ связи.
Сегодня на отводах сети ВЧ связь рассматривается как система, которая надежно передает данные систем защиты и обеспечивают прозрачный удобный интерфейс для данных и речи от широкополосных цифровых сетей до конечного потребителя при значительно большей пропускной способностью, по сравнению с обычными аналоговыми
38
системами. С современной точки зрения высокая пропускная способность может быть достигнута только путем увеличения полосы частот. То, что в прошлом было невозможно из- за недостатка свободных частот, сегодня реализуется благодаря повсеместному применению оптических линий. Поэтому ВЧ системы усиленно используются только на ответвлениях сети. Также существуют варианты, когда отдельные участки сетей объединены между собой ВОЛС, что позволяет использовать одинаковые рабочие частоты гораздо чаще, чем в случае объединенных систем ВЧ связи.
Всовременных цифровых ВЧ системах плотность информации при использовании быстрых сигнальных процессоров и цифровых способов модуляции может быть увеличена по сравнению с аналоговыми системами с 0,3 до 8 бит/сек/Гц. Таким образом, для полосы частот 8 кГц в каждом направлении (прием и передача) может быть достигнута скорость 64 кбит/с.
В2005 году фирма Siemens представила новую цифровую аппаратуру ВЧ связи «PowerLink», подтвердив лидирующее положение в данной области. Аппаратура PowerLink сертифицирована и для использования в России. Создавая PowerLink фирма Siemens создала мультисервисную платформу, пригодную как для аналогового, так и для цифрового применения.
Рис. 2
Уникальные особенности этой системы.
Оптимальное использование выделенной частоты: лучшая аппаратура ВЧ связи позволяют передавать данные со скоростью 64 кбит/с и менее, в то время как у PowerLink данный показатель составляет 76,8 кбит в секунду, занимая полосу 8 кГц.
Больше речевых каналов: еще одной инновацией фирмы Siemens, реализованной в системе PowerLink, является возможность передачи 3-х аналоговых речевых каналов при полосе 8 кГц вместо 2-х каналов в обычной аппаратуре.
Видеонаблюдение: PowerLink — первая система ВЧ связи позволяющая передавать сигнал видеонаблюдения.
AXC (Automatic Crasstalk Canceller) Автоматическое подавление перекрестных помех: раньше сближенные полосы приема-передачи требовали сложную ВЧ настройку для
39
минимизации влияния передатчика на свой приемник. Запатентованный AXC блок заменил сложную гибридную настройку и соответствующий модуль, а качество приема-передачи улучшилось.
OSA (Optimized Sub channel Allocation) Оптимальное распределение подканалов:
еще одно запатентованное решение компании Siemens гарантирует оптимальное распределение ресурсов при конфигурировании услуг (Речь, данные, защитная сигнализация) в выделенной частотной полосе. В результате итоговая передающая емкость увеличивается до 50%.
Повышенная гибкость: для обеспечения надежности инвестиций и возможности будущего использования фирма Siemens реализовала функцию «ease-up!» для простого и надежного обновления.
Многофункциональное оборудование: выполняя проект на базе комбинированной аппаратуры PowerLink вы можете забыть об ограничениях которые были в обычных терминалах при планировании частот. С PowerLink Вы сможете спроектировать систему ВЧ связи со всем набором услуг (передача речи, данных, сигналов РЗ и ПА) в доступной полосе. Один комплект PowerLink может заменить три (3) обычных аналоговые системы.
Рис. 3
Литература
1.Energie Spektrum, 04/2005: S. Schlattmann, R. Stoklasek; Digital-Revival von PowerLine.
2.PEI, 01/2004: S. Green; Communication Innovation. Asian Electricity 02/2004: Powerline Carrier for HV Networtk.
3.Middle East Electricity, Feb. 2003: J. Buerger: Transmission Possible.
4.Die Welt, April 2001; J. Buerger: Daten vom Netz ubers Netz.
5.VDI Nachrichten 41; Oktober; 2000 M. Wohlgenannt: Stromnetz ubertrugt Daten zur eigenen Steuerung. Elektrie Berlin 54 (2000) 5-6; J. Buerger, G. Kling, S. Schlattmann: Power Line Communication-Datenubertragung auf dem Stromverteilnetz.
6.EV Report, Marz 2000: J. Buerger, G. Kling, S. Schlattmann: Kommunikationsruckrat fur Verteilnetze.
7.ETZ 5/2000; G. Kling: Power Line Communication Technik fur den deregulierten Markt
8.http://market.elec.ru/nomer/15/systems-vch/.
40