Contents Telektronikk - Telenor
Contents Telektronikk - Telenor
Contents Telektronikk - Telenor
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
78<br />
with two to four hours read-out, can be<br />
proper. The dimensioning can then follow<br />
Moe’s principle.<br />
Many of the overflow networks are very<br />
complicated, with lots of alternatives.<br />
The complexity of the network may be<br />
limited by the risk of jeopardising its<br />
manageability, the traffic control and the<br />
means of measuring. If these are exceeded,<br />
the complication can become<br />
problematic.<br />
One may ask if these observations and<br />
conclusions, mainly dating from the<br />
1980s, are pertinent in the future network<br />
services: now the components are cheap<br />
and capacity costs almost nothing. A few<br />
comments will be given.<br />
Service quality of a network depends on<br />
some bottlenecks in the network. These<br />
impair the service when the load limit is<br />
reached. The bottlenecks are probably<br />
not in transmission routes any more, but<br />
in signalling, data transmission channels<br />
or processors, getting blocked first.<br />
These limits must be identified by the<br />
operator, who has to know how near the<br />
calculated limits the actual loads are.<br />
There, the peak-hour or other dimensioning<br />
period intensities with their models<br />
will be needed for capacity calculations,<br />
even in the future. Some of the load variations<br />
follow these, as described above,<br />
some are of new types. Measurements<br />
clear the traffical facts – models just help<br />
in thinking. The augmentations can then<br />
be prepared in time. Otherwise, the operator<br />
will be in for a surprise.<br />
Network economy becomes more complicated<br />
when, instead of circuit-groups, the<br />
dimensioning objects are clusters, networks,<br />
and groups of competing operators’<br />
networks, serving clients of various<br />
services. This higher level of traffic<br />
dimensioning and optimising is not easy<br />
to manage without knowledge of the<br />
simple components’ traffical relations.<br />
Or what is even worse, if the complicated<br />
network is still controlled on the basis of<br />
obsolete elementary conceptions.<br />
Traffic modelling traditionally concentrates<br />
on Poissonian distribution inside<br />
the hour and regular variations between<br />
the hours, days and seasons. The above<br />
described measurements do not support<br />
this concept: traffic variations usually<br />
exceed the expected. Rather some grade<br />
of self-similarity, known in mathematics<br />
of fractions, can be found, which opens<br />
ways to very different thinking in the<br />
future.<br />
13 Acknowledgements<br />
The above described measurement have<br />
been made during several years, partly to<br />
answer some questions raised by the<br />
CCITT Study Group 2, partly arisen from<br />
local interest. I owe my warm thanks to<br />
the following specialists and colleagues<br />
at Helsinki Telephone Company: Dr.<br />
Pekka Lehtinen programmed the<br />
AUTRAX equipment to produce several<br />
traffic data simultaneously and continuously,<br />
and analysed traffic variations.<br />
MSc Matti Mäkelä and BSc Kari Laaksonen<br />
with their staff carried out the traffic<br />
measurements in practice. MSc Yrjö<br />
Viinikka of Oy Comptel Ab took care of<br />
the data handling.<br />
14 Documentation<br />
The above referred studies about optimising<br />
and the traffic measurements in<br />
Helsinki, have been published in full in<br />
the following papers:<br />
1 Ahlstedt, B V M. Load variance and<br />
blocking in automatic telephone traffic:<br />
load duration, a basis for dimensioning<br />
: blocking duration, a yardstick<br />
for traffic quality. I: 5th international<br />
teletraffic congress, ITC 5,<br />
524–530, New York, 1967.<br />
2 Lehtinen, P I. Extreme value control<br />
and additive seasonal moving average<br />
model for the evaluation of daily<br />
hour traffic data. I: 9th international<br />
teletraffic congress, ITC 9, Torremolinos,<br />
1979.<br />
3 Parviala, A. The suboptimal alternate<br />
routing practice with non-coincident<br />
busy hours. I: 10th international teletraffic<br />
congress, ITC 10, Montreal,<br />
1983.<br />
4 Parviala, A. The dimensioning of<br />
alternate routing by a simple algorithm,<br />
taking into account the nonideal<br />
parameters. I: 5th Nordic teletraffic<br />
seminar, NTS 5, Trondheim,<br />
1984.<br />
5 Parviala, A. The stability of telephone<br />
traffic intensity profiles and its<br />
influence on measurement schedules<br />
and dimensioning. I: 11th international<br />
teletraffic congress, ITC 11,<br />
Kyoto, 1985.<br />
6 Parviala, A. The stability of daily<br />
telephone traffic intensity profiles<br />
and its influence on measurement<br />
routines. I: 6th Nordic teletraffic<br />
seminar, NTS 6, Copenhagen, 1986,<br />
and Sähkö 60, 26–30, 1987.<br />
7 Parviala, A. Peak-hour and other<br />
peaks of day. I: 8th Nordic teletraffic<br />
seminar, NTS 8, Espoo, 1987.<br />
8 Parviala, A. The stability of the daily<br />
intensity profiles and its influence on<br />
the choice of measurement routines<br />
in telephone traffic in single lowcongestion<br />
circuit-groups or overflow<br />
clusters. I: 12th international<br />
teletraffic congress, ITC 12, Torino,<br />
1988.<br />
9 Parviala, A. The reality of busy hour<br />
in history and in measurements. I:<br />
9th Nordic teletraffic seminar, NTS<br />
9, Kjeller, 1990, and I: 13th international<br />
teletraffic congress, ITC13,<br />
Copenhagen, 1991, WS History.<br />
10 Parviala, A. The year’s highest vs.<br />
consecutive days. I: 10th Nordic teletraffic<br />
seminar, NTS 10, Aarhus,<br />
1992.<br />
11 Parviala, A. Traffic reference period.<br />
I: 11th Nordic teletraffic seminar,<br />
NTS 11, Stockholm, 1993, and I:<br />
14th international teletraffic congress,<br />
ITC 14, Antibes, 1994. Spec.<br />
Traffic Measurements.