Multiple applications, such as mobile communications regulatory projects, education, dimension- ing and optimization of mobile networks, spectrum impact on costs analysis, etc., are suitable to be solved by the dimensioning and optimization algorithms detailed in Chapters3 and 4.
In this chapter, the performance, efficiency and application of the developed dimensioning and optimization algorithms are evaluated by means of their application to two different experiments: • Experiment A: Techno-economical study based on the efficiency comparison between the HSPA and LTE technology to provide the Mobile Broadband Access Service. The goal of this case is to determine the most suitable technology, HSPA or LTE, to provide the internet broadband service based on the required investment costs of the cellular network deployed. Only the network dimensioning module is applied, no optimization algorithms are used to balance the traffic demand over the technologies. The distribution of the traffic demand over the set of technologies considered is defined by the user.
• Experiment B: Performance evaluation of the metaheuristic optimization algorithms. The aim of this experiment is to evaluate the impact on the resulting investment costs of an optimal service distribution over the set of technologies to be deployed. The developed optimization algorithms, EA, CRO and TLBO, are applied to obtain the most efficient user demand distribution, in terms of investment costs, to be used by the cell dimensioning module to carry out the estimation of the total network equipment. Finally, the investment costs associated to the optimal service distribution and the investment cost of a experience- based traffic distribution are compared.
6.1
Spanish country definition
Experiments are focused on the spanish country. Information about the geographical areas of Spain is obtain from the data provided by MapInfo. MapInfo is a geographic information system software, and it divides the country of Spain into ten thousand geographical areas. The first step is the generation of the set of cities and villages where the radio access network is going to be deployed. In order to achieve this goal, the data provided by MapInfo is used by a Scenario Generator module (ScenGen). The ScenGen defines a set of aggregator areas (AgA). The rest of geographical areas (GeA) are aggregated to these aggregator areas in an iterative process. This aggregation process is based on two different set of thresholds, population density and distance thresholds for each type of area (urban, suburban and rural). The process is summarized as follows:
• Data obtained from MapInfo is sorted by the type of area. Urban areas are at the top, suburban areas in the middle, and rural areas are at the bottom.
• For each type of area, they are ordered according to population density.
• The GeA located at the top is chosen as the first AgA. In the following iteration, the AgA is the one with the highest population density, among the not yet aggregated areas. • Once the AgA is identified, a threshold comparison phase starts. The process evaluates the
population density of the AgA. In case that the population density is larger the population density threshold defined for that type of area (urban, suburban and rural), all the not yet aggregated geographical areas inside the aggregation radius defined are aggregated to the chosen AgA.
This process defines the cities file where the radio access network is going to be deployed. In our case, it defines a set of 5354 different cities. For each of the cities the most relevant information, among other parameters, is:
• Total population. Urban, suburban and rural population is detailed.
• Total extension of the city, together with the corresponding area in square kilometers for urban, suburban and rural type of areas.
• Type of terrain, flat, hilly and mountainous. • Average height of buildings.
• Type of radio access network deployment for every area within a city.
Once the cities are configured, this file is included in the scenario E, and the simulations of the experiments are launched. Note that it is also required an estimation process to carry out the user demand of the different services defined in the service profile. This process is described in Section6.2.
6.2
Traffic load definition process
The input data used for all the scenarios considered is based on data obtained from the Spanish Telecommunications Market Commission annual report [109]. First, the traffic load of the voice service is obtained from the total annual minutes consumed by the subscribers, Θpy. In this work, a month is considered to have 22 business days, Dm, and 6 business hours per day, Hd.
Based on this data, and assuming one call attempt per busy hour, the traffic load of the voice service is estimated as indicated in Equation6.1.
avoice[mE] =
Θpy/(12 · Dm· Hd)
60 (6.1)
The traffic load of the data services, as there is no information in the Spanish Telecommuni- cations Market Commission (CMT) report, is treated as voice-equivalent miliErlangs1 (ve-mE), and therefore obtained from the value estimated in6.1 as indicated in Equation6.2.
ai[ve-mE] = avoice·
Rbj
Rbi
, ∀i 6= j ∈S (6.2)
where Rbj is the binary rate of the voice service, and Rbi corresponds to the binary rate of the
data service i, defined in the input service profile.
Finally, the traffic load of the short message service (SMS) and the multimedia message service (MMS) is calculated. The traffic load of the SMS is calculated from data in [109]. The starting point is the number of SMS per user per year, Msms, and based on the characteristics of this service, 128 bytes of length, Lm, the traffic per user is estimated.
asms[mE] = Msms 12 · Dm· Hd · Lm· 8 Rbsms (6.3) The MMS traffic load is estimated as a percentage of the SMS traffic, this percentage is based on the existing ratio, r, between the SMS and MMS for the whole set of operators in Spain, see [109].
amms[mE] = asms· r (6.4)
In case of the MBAS, the throughput per user is most relevant parameter when dimensioning the radio access network. It determines the total volume of downloaded data. Taking as reference the value in the CMT report, the total volume of downloaded data in year 2011 was of 90.500 Terabytes. The total population registered in Spain in the last 2011 was of 46.152.926 inhabitants, based on this value the corresponding volume of downloaded data per user per month, , Vu, is estimated. This value is taken as reference in order to calculate the minimum guaranteed binary rate for the MBAS that is used to dimension the HSPA and/or LTE network, parameters such as considered business days per month, and hours per business day are taken into consideration in this calculation.
Rb[M bps] =
(Vu· 8)/(Dm· Hd)
3600 (6.5)
The process described is required to obtain the user demanded traffic for every experiment defined. The final results of the amount of traffic per service and the total demand is shown in tables in the corresponding experiment definition’s sections.