• No results found

System simulation and simulator features

3. Dynamic Resource Allocation for OFDMA

3.3. System simulation and simulator features

The optimization of cellular networks can be performed on many different levels and for a large number of different aspects. This includes static optimizations during the network planning process, as well as highly dynamic optimization during network operation. The latter, the process being increasingly automated through SON. Here, we investigate DRA performance and optimization at the planning stage by using system level simulations, since we are interested in the overall performance of the

49

network, following the typical KPI conditions, namely the supported average data rate of users in a certain area, network QoS, among others.

In order to reduce the complexity, usually it is advantageous to decouple simulators working at different levels of abstraction. In this case, appropriate interfaces are required between the simulators, namely, system level simulators and link level simulators. Some problems, however, may require the combination of different simulation levels in one simulator. The specific case of interfacing the system level simulation with link level is evaluated in the section 3.4.1.3.

In this section we describe the methodology for system level simulation incorporating DRA metrics. We focus on OFDMA cellular and typical environments namely urban, including accurate radio channel and interference modeling.

3.3.1.1 Simulator features

The complexity of existing and future cellular mobile networks makes their optimization and performance evaluation a challenging task. Although many aspects of the cellular networks can be evaluated analytically, the use of testbeds or computer simulations are often the only way to be able to acquire performance results closer to real scenarios. In many cases, especially in the design phase of new standards or for dynamic algorithms optimization phase, computer simulations are a commonly applied method, since it provides an easier implementation and maintenance solution, when compared to testbeds.

Based on this purpose, a collaborative work resulted in an in-house system level simulator for cellular systems that was designed and implemented initially to obtain

50

performance results closer to real scenario of one of the access technologies candidates of what was supposed to be the 4G of mobile communication systems, based on Multi-Carrier Code Division Multiple Access (MC-CDMA) [78]. The design was proposed to be modular to facilitate both collaborative work and also adaptive to different cellular data packet standards and requirements. This modular PHY-MAC based dynamic system simulator has evolved since then for UMTS HSPA [53] [79] and lately to LTE standard. Since relevant standardization bodies and industry consortia have specified reference scenarios, model parameterizations, and metrics, they are used in order to allow a comparison of simulation results generated by different research groups. These simulation specifications can be found in [80] for the 3GPP and in [81] for the Next Generation Mobile Networks (NGMN) Alliance.

A part of the fact that is an in-house simulator tool, with inherent flexibility and support to the authors, this simulation tool is fully developed in C++, under Linux operating system, in opposite for example to other simulators based in other platforms [82], which makes it suitable for heavy and complex processing requirements of the multi-cell, multi-user MAC algorithms of dynamic cellular systems.

3.3.1.2 Blocks developed for the purpose of this thesis

This chapter describes methodology used for system level simulator used in this work. The methodology focuses on OFDMA cellular and typical environments namely urban, including accurate radio channel and interference modeling.

51

A part of the detailed description of the simulator and its features, in this section we indicate particular extra modules that were developed and included in the simulator described in this chapter, for the sole purpose of the work presented in this thesis.

SON module

The multiple cells approach is shown in Figure 3.5, where in each cell there is one base station and a random number of mobiles, randomly located inside each cell. In the conventional cellular system, each cell independently manages its resources through MAC, as described in the section Figure 3.1. For SON based implementation, a centralized approach is need in order an cells independent entity be able to manage cells resources, or at least collect information and give to the cell management directives. For this purpose a SON module was implemented in the simulator and interacts with all cells. The two approaches, with independent cell resource managements and SON based resource management are presented in Figure 3.2 and Figure 3.3 respectively. SON based management is carried-out in the chapter 4.

Figure 3.2: Conventional Wireless Cellular System with multiple cells, with independent resource management.

52

Figure 3.3: SON based resource management Cellular System.

Manhathan Scenario for Coordinated Multi-Point

As described, the simulator was designed for cellular systems and all deployment follows the cellular model of Figure 3.5, a new deployment was proposed and implemented for CoMP studies of chapter 5. based on Manhathan model. Manhattan Model [83] is proposed in [84] for broadband systems and is an attractive scenario for deployment of the CoMP in a dense urban environment, where fiber has already been deployed and can be reused to transport the radio signals to/from the RAUs, or can be economically deployed. Furthermore, end users will be mainly in line-of-sight conditions or near line-of-sight with the antenna. This Manhattan deployment consists of a rectangular regular grid of RAUs. This rectangular grid follows the regular street structure of a city with buildings, streets, and blocks of the same size. The RAUs are considered to be deployed at optimal positions on the streets (see Fig. 2).

53

CoMP Area 1 CoMP Area 2 CoMP Area N

s b RRH Building s= Street Width b = Building Width User d d1 d2 LOS NLOS LOS

NLOS Non Line-Of-Sight

Line-Of-Sight

Figure 3.4: Hexagonal distributed deployment vs Manhattan network deployment with joint processing areas.

Related documents