Sascha Berger

Joint Downlink and Uplink Tilt-Based Self-Organization of Coverage and Capacity Under Sparse System Knowledge

Concurrent self-organization of coverage and capacity in both downlink (DL) and uplink (UL) is an ambitious task since 1) Coverage and capacity are conflicting key performance indicators (KPIs), and 2) an adequate DL performance does not necessarily entail an adequate UL performance because the DL and UL interference scenarios are fundamentally different. However, considering the UL when self-organizing the network is crucial because the UL transmission becomes more important due to the emergence of new applications and services, such as social networking, video calls, or sensor networks, which require either parity between DL and UL traffic or even more UL than DL traffic. In this paper, we propose a general concept for the self-organization of multiple KPIs while having only very sparse knowledge about the network. Using this concept, we propose an effective tilt-based algorithm that manages to jointly optimize coverage and capacity in DL and UL. In an urban scenario with real Long-Term Evolution (LTE) site locations, we were able to increase the cell-edge user throughput by 70% (DL) and 24% (UL) while concurrently decreasing the number of uncovered users by 21% (DL) and 25% (UL), compared with a well-chosen reference tilt setting.

Comparing Online and Offline SON Solutions for Concurrent Capacity and Coverage Optimization

Self-organizing networks (SONs) can carry out their optimization procedures in an on- or off-line manner. On one hand, an online SON solution optimizes network parameters during operation. On the other hand, an offline SON solution employs a simulation environment of the network to be optimized in order to perform an offline parameter optimization before applying changes to the network. Thus far, researchers have not yet compared the characteristics of on- and offline SON solutions and they typically do not comment their SON solutions operational mode. However, specifying the SONs operational mode is crucial because it determines the type and number of measurements to be performed, and it decides whether it is required to accurately model the network to be optimized or not. In this work, we compare the general properties of on and off-line SON solutions qualitatively and compare an on and an off-line algorithm for coverage and capacity optimization quantitatively, using a realistic LTE simulation scenario. Based on the results obtained, we can conclude that offline SON solutions should be preferred as long as the required inputs are available. However, online SON solutions provide an adequate alternative to offline SON solutions if some of the inputs required are missing.

Online Antenna Tilt-Based Capacity and Coverage Optimization

In the field of self-organizing networks (SONs), the use case of concurrent capacity and coverage optimization (CCO) is challenging since the objectives are typically conflicting and adequate solutions often require interaction between a multitude of network sites. As a consequence, solving the use case of CCO is particularly difficult. This letter proposes an efficacious modification of the antenna tilt-based SON algorithm by Berger et al. We propose using a probability distribution function (PDF) of throughput measurements and an estimate of the number of covered and uncovered users of each cell considered for optimization in order to construct an objective function that is better suited to jointly maximize throughput and coverage. Due to the limited amount of data required for computing the algorithms objective function, the SON solution proposed can be implemented not only following a centralized but also following a distributed architecture. Simulations show that the SON algorithm can exploit this additional information to increase its performance in terms of throughput gains.

Modelling the impact of downlink CoMP in a realistic scenario

Coordinated multi-point transmission/reception (CoMP) enables the communication between different base stations, which can improve the cell edge user performance as well as the system capacity. So far, the most studies focused on evaluating the performance of CoMP in regular scenarios with uniform traffic. Since these scenarios differ substantially from realistic conditions, it is doubtful whether the performance results obtained in these studies can be applied to real-word scenarios. That is why in this work we evaluate the performance of various CoMP methods for a European city. As we use the existing network topology, employing a high resolution ray-tracing path loss prediction and adopting a traffic map from live 3G network, we examine a highly realistic scenario. The results show moderate overall capacity gains but large reduction in network outage of advanced data services.

On the advantages of location resolved input data for throughput optimization algorithms in self-organizing wireless networks

Performing and evaluating Call Trace (CT) or also called Minimization of Drive Test (MDT) measurements entails additional expenses to the operator as well as to the users. However, by using location specific network data obtained from the measurements as additional information for self-organization procedures, we expect these procedures to perform essentially better since they can take their decisions upon an improved knowledge base. In this paper, we investigate the performance gain induced by the use of location specific input data for self-organization procedures. For this purpose we propose, evaluate, and compare two force field based algorithms, which are dedicated to optimizing the coverage as well as the user throughput in a macro cell deployment with heterogeneous traffic demand. The results of this work confirm the expectations: It is shown that using the additional location specific information the self-organization procedure investigated leads to higher average throughput gains and an improved robustness in the sense that the procedure was able to successfully handle all scenarios investigated.

Force field based joint optimization of strictly monotonic KPIs in wireless networks

Self-organizing networks (SONs) are expected to improve the quality of service (QoS) while reducing the operators capital and operational expenditures. In order to achieve this ambitious goal, SON concepts are required to manage multiple key performance indicators (KPIs) by modifying several different network control parameters. Developing concepts and algorithms that cope with such complexity is not easy. In order to avoid the complexity of joint optimization, the research so far focused on concepts, which coordinate multiple algorithms [1], [2]. Usually, each of these algorithms is dedicated to managing a single use case or KPI. In contrast, we introduce and analyse a force field based joint optimization concept which considers all KPIs and network control parameters concurrently. Despite the holistic approach, our concept offers an acceptable complexity. Using an algorithm based on this concept, we can resolve a high load scenario by jointly adjusting the tilts and cell individual offsets (CIOs).

What is the advantage of cooperation in self-organizing networks?

Self-organizing network (SON) functions can be characterized by their required cooperation between the network elements (NEs). The cooperation among the NEs can include a multitude of possible actions, such as reporting of alarms or coordination of joint parameter modifications at multiple NEs. However, the question for the advantage of cooperation among the NEs in SONs is still an open research topic. By limiting the cooperation between the NEs, the required architectures for utilizing the SON function at hand can be simplified, which in turn can lead to cost savings. In this paper, we investigate the impact of degraded cooperation among the NEs on the SON architecture required and on the performance in a joint capacity and coverage optimization (CCO) use case. For the scenario investigated, we observe that the performance decreases dramatically when decreasing the cooperation among the NEs. However, we can also show that the exchange of information, such as the values of considered key performance indicators (KPIs), among the NEs is more important for an efficient operation than the coordination of the NEs actions. Our results show that, a centralized approach outperforms distributed and localized approaches for the CCO use case investigated.

Energy efficiency optimization for 2D antenna arrays in self-organizing wireless networks

Nowadays telecommunication systems consume a huge amount of energy. While energy efficiency methods have already been studied for linear antennas, this paper focuses on 2D antenna arrays. An antenna array with many elements gives the network operator a high degree of flexibility in network optimization, which can be used not only for improving the quality of service, but also for increasing the energy efficiency. The energy efficiency can be improved by adapting the array shape, i.e., by switching off certain elements of the array. This work presents three novel solutions for the energy efficiency optimization, which are based on Q-learning, biological neural networks, and random drop, respectively. Simulation results demonstrate that significant improvements in terms of energy efficiency can be achieved, while the 5th percentile of user throughput and coverage performances decrease only marginally.