Mobile communication techologies develop in an intense way nowadays. The bandwidth of channels grows and the number of services increases. Looking for 20 years in the past there was only radio and TV, but today almost everyone owes multiservice gadget that allows him to make phone calls as well as provides an Internet access and allows to make videocalls and teleconferences. Besides, mobile networks have plenty of applications from building home data networks to global positioning.

The LTE technology is one of the most perspective 4G mobile technologies. But the deployment of the 4th generation network could make some troubles. First of all the free wide frequency channel should be allowed. Also, the requirements for the core network are higher. Setting of the new equipment requires a solid amount of investments.

Using the self-organising network (SON) concept, the deployment of the new network can be done less expensively. The SON concept if intended for the network operation optimization. The main idea of SON is reducing OPEX. This is done by substitution of manual work by automated processes with the help of special software. The quality of the work increases as well. Reducing the number of employees results in reducing the OPEX by about 30%.

The three main principles of the SON concept are self-configuration, self-optimization and self-recovery.

The aim of this thesis is increasing of LTE network efficiency along with the requirements for the QoS parameters satisfaction. This should be done by effective frequency and technical resourses usage.

A number of tasks should be solved to reach the aim:

1. The research of the current mobile market and analysing the perspectives of 4G network
2. Study the efficiency of SON solutions applied to LTE network
3. Develop the criterion of provider resources usage efficiency
4. Develop the mathematical and software model of dynamic resources reallocation
5. Research the developed model and give recommendations for mobile provider considering technical solutions

The project has the following scientific innovation:

• The integral criterion of resource consumption is developed
• The efficiency of LTE/SON solution is proven
• The mathematical model of the self-organising network is developed with the use of ant colony optimization algorithms
• The software that implements the mathematical model is developed
• Technical recommendations are given for mobile provider about the network evolution to LTE

The aim of the SON concept is the mobile network optimization. Obviously, the criterion of efficiency should be developed. The work aims to maximize the provider's profit. Thus the bandwidth provided must me as close to the bandwidth declared as possible. It must be considered that revenue per user (RPU) depends nonlinearly on the bandwidth provided.

Mathematically the criterion is as follows:

where Y_ik – bandwidth provided for i^th user of k^th category, C_k(Y) – RPU value of k^th category for particular bandwidth provided, C_k.max – maximum possible RPU for this user, Y_ik.required – bandwidth that requires i^th user.

Virtually, to maximize the value of K capacity loss must be minimized. In an existing mobile networks (for example GSM) it is done by decreasing the bandwidth for low-priority users. SON concept allows to optimize by reallocating user to base station (BS) connections. Obviously it does not always solve the problem. So, this approach must be combined with the bandwidth decreasing.

Usually the methods of teletraffic theory are used to build a mathematical model of mobile network. But there is a difference in LTE/SON simulation. In particular, some BS parameters could change their values during SON operation. These processes are very similar to neural networks therefore a BS could be represented as a neuron. Considering this, it is effective to apply the methods of mass population behaviour simulation. In this case users represent members of this population. The most effective way is to use ant colony optimization (ACO) algorithms. This model considers possibility of "path labeling" with pheromone that provides feedback.

So optimization in the developed model occurs in 2 stages: users redistribution by BS and possible bandwidth reduction for low priority users.

Figure 1 - The general algorithm of the optimization process

In LTE each base station has two service areas - primary, in which the signal from this station far exceeds capacity of all other signals, and peripheral, in which subscribers can be connected to one of several base stations. For users who are in the peripheral zone, BS selection algorithm is based on the array of base stations priorities, that are calculated by the formula:

where P_j – j^th BS priority for current user, d_j – distance from j^th BS priority to current user, K_ho – handover coefficient.

The subscriber is passed to the BS with the highest priority. After calculating the priorities of all the BS and appointment of each subscriber to a certain BS, the load that each BS requires from the transportation network is calculated. In fact, the data about the location of subscribers is transformed to data about the distribution of load on BS.

The described model is static, ie considers only one subscribers and base stations distribution. It implements the SON function that is called Mobility load balancing (MLB). But in real systems, mobile subscribers are always moving, and therefore there are additional problems that need solutions, including the problem of accurate handover. There is another function of SON that allows to control handover - Mobility Robustness Optimization (MRO). It is implemented in model by handover coefficient in BS priority formula.

The value of this coefficient equals 1 if the considered BS is the BS that serves current user. Otherwise, its value lies in the range (0;1). The negative impact of handover effect for user is a 30 ms delay that user can hardly percept, but the load that is created in a network should be considered. This operation involves two base stations, so in the model we can assume that the user that makes handover is served by both BS during a sample interval. This assumption will take into account the negative handover impact on network performance.

The abovementioned model is implemented in software. The programming language is Ruby.

The area in Donetsk was chosen for simulation with defined location of base stations and random loacation of subscribers considering the usual location of subscribers of every type. The aim was to develop recommendations for the provider about an optimum number of base stations and frequency band using the traditional approach to network deployment and SON approach.

Figures 2a and 2b represent charts that show the dependence of efficiency criterion value from the number of base stations and frequency band, with a fixed number of subscribers. Using the charts, it could be recommended to build the network using the SON software with the number of base stations of 75 per area. Required frequency resource is 10 MHz. This ensures provider's profit around 98% of the possible for the number of subscribers that was selecter for simulation.

a) traditional approach   b)SON approach
Figure 2 - K(Nbs,ΔF) dependence

The K(Nusr,Nbs,ΔF) dependence, where N_usr - the number of subscribers, N_bs - the number of base stations, ΔF - frequency band. Charts for this dependence are given in figures 3a and 3b. The value of criterion K represented by color.

a) traditional approach   b)SON approach
Figure 3 - K(Nusr,Nbs,∆F) dependence
This is an animation. To restart it please refresh the page.

Since the criterion value is proportional to the providers profit, it can be used to determine the optimal network configuration comparing its value with a profit from the network. The increasing number of subscribers can also be considered this way.

These charts give real information to the provider on the number of base stations that must be placed in an area, the required frequency band and the number of subscribers in the area, that are located by the specified probability density distribution that should be served with the specified quality of service. The simulation can be performed for any area, but it is worth noting that the optimization task in this case is NP-complete, and although the metaheuristic algorithm is used and the amount of computation is reduced, but computation time grows exponentially with the linear base stations number growth.

The masters thesis is not finished to the time of writing this abstract. The research is expected to be finished till Decemeber the 1^st, 2011. Contact author or scientific advisor after this date to obtain the full text of the thesis.

1. И.В. Дегтяренко, Д.С. Шахов, Д.А. Кнерцер, А.А. Орехов Модель розвитку мережі мобільного оператора при використанні технології LTE/SON // Наукові праці Донецького національного технічного університету. Серія: Обчислювальна техніка та автоматизація. Випуск: 20 (182). Стр. 130-136.
2. Кнерцер Д.А., Дегтяренко И.В. Розробка та дослідження моделі cегмента мережі, що самоорганізується, на базі технології LTE // Тези всеукраїнського конкурсу студентських наукових робіт з природничих, технічних та гуманітарних наук у 2010/2011 р.р.("Телекомунікаційні системи та мережі", "Інформаційні мережі зв'язку"). - Одеса: ВМВ, 2011. - 24 с.
3. Moray Rumney. LTE and the Evolution to 4G Wireless: Design and Measurement Challenges. - Agilent Technologies. – 557 p.
4. Farooq Khan. LTE for 4G Mobile Broadband: Air Interface Technologies and Performance. – Cambridge University Press. – 492 p.
5. Стандарт ETSI TR 136 902 V9.3.1 (2011-05). LTE; Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Self-configuring and self-optimizing network (SON) use cases and solutions
6. LTE SON: How Does It Work? [Електронний ресурс] - Режим доступу http://wwwen.zte.com.cn/endata/magazine/ztetechnologies/2009year/no5/articles/200905/t20090518_171482.html вільний. Назва з екрана.
7. Opportunities with self organizing / self optimizing networks (SON) [Електронний ресурс] - Режим доступу http://radioplan.us/docs/oliver_wyman_actix_son_analysis_for_website.pdf вільний. Назва з екрана.
8. Marco Dorigo, Thomas Stuzle. Ant Colony Optimization. – Massachusetts Institute of Technology. – 2004. – 368 p.
9. Муравьиные алгоритмы [Електронний ресурс] - Режим доступу http://habrahabr.ru/blogs/algorithm/105302/ вільний. Назва з екрана.
10. 3GPP Long-Term Evolution / System Architecture Evolution Overview [Електронний ресурс] - Режим доступу http://www.ikr.uni-stuttgart.de/Content/itg/fg524/Meetings/2006-09-29-Ulm/01-3GPP_LTE-SAE_Overview_Sep06.pdf вільний. Назва з екрана.