Peter Szilagyi

An Automatic Detection and Diagnosis Framework for Mobile Communication Systems

As the complexity of commercial cellular networks grows, there is an increasing need for automated methods detecting and diagnosing cells not only in complete outage but with degraded performance as well. Root cause analysis of the detected anomalies can be tedious and currently carried out mostly manually if at all; in most practical cases, operators simply reset problematic cells. In this paper, a novel integrated detection and diagnosis framework is presented that can identify anomalies and find the most probable root cause of not only severe problems but even smaller degradations as well. Detecting an anomaly is based on monitoring radio measurements and other performance indicators and comparing them to their usual behavior captured by profiles, which are also automatically built without the need for thresholding or manual calibration. Diagnosis is based on reports of previous fault cases by identifying and learning their characteristic impact on different performance indicators. The designed framework has been evaluated with proof-of-concept simulations including artificial faults in an LTE system. Results show the feasibility of the framework for providing the correct root cause of anomalies and possibly ranking the problems by their severity.

Network side lightweight and scalable YouTube QoE estimation

YouTube is a prominent Over The Top (OTT) application increasingly used on mobile devices for watching online videos. Due to the mobile environment, cellular technology specific reasons and the delivery mechanism adopted by YouTube, providing seamless customer experience is not straightforward. Being a popular application whose availability is important from customer satisfaction point of view, operators must deploy mechanisms capable of network wide YouTube QoE measurement and degradation detection. Known solutions aim to provide a comprehensive set of QoE influence factors such as the exact stallings for each video session. However, in order to generate network wide insight to a statistically relevant set of all YouTube downloads, such mechanisms may not represent an optimal balance between the provided level of details and their cost or complexity. Instead, this paper proposes a lightweight, real time, network side QoE estimation method suitable for network wide deployment. The proposed solution is able to detect YouTube videos with QoE degradations through novel metrics that were validated with real network measurements. The paper also presents the statistical analysis of a large set of videos showing how the method can evaluate the YouTube QoE in a network.

Enhanced Mobility Load Balancing Optimisation in LTE

Self-Organising Network (SON) use cases are seen as key enablers of efficient Long Term Evolution (LTE) system operation. Mobility Load Balancing Optimisation (LB) is a prominent SON use case defined by 3GPP with the scope of balancing the load in a given coverage area by reconfiguring the handover thresholds. When overload is detected at a given cell, the LB reduces the handover thresholds of those user equipments that can be redirected to the underutilised neighbouring cells. This mechanism improves the user plane performance in the overloaded cell. Despite the potential benefits, the available LB mechanisms have two major deficiencies, both of them causing user experience degradation: on the one hand, they tend to overestimate the load on the air interface, resulting in unnecessary handovers; on the other hand, they are either transport agnostic or have limited capability to consider the transport load during their decision process. This paper provides an overview of the LB related problems and proposes an improved solution including accurate evaluation of the radio load and a transport aware mechanism for end-to-end optimisation. The performance of the solution was evaluated by simulations.

Radio Channel Degradation Detection and Diagnosis Based on Statistical Analysis

In complex radio environments like cellular mobile networks, the quality of the radio channel strongly affects the service quality experienced by users. Degraded channel quality makes users unsatisfied and thus results in revenue loss for the operator. This is particularly important as most of the cell level faults result in channel degradations. As the complexity of commercial cellular networks is increasing, the need for reliable automated detection of cells with degraded radio quality has become essential. Channel degradation detection alone is already a challenging task but finding the root cause is usually even harder. Human experts with decades of experience in the telecommunication domain are having hard times during investigating the root cause of degraded service quality. Our demonstration introduces a framework that is on one hand capable to detect cells with channel degradations and on the other hand differentiate between possible root causes. The showcases are implemented in a simulated 3GPP Long Term Evolution (LTE) environment animated on a 3D graphical scene.

Physical Cell ID Allocation in Multi-layer, Multi-vendor LTE Networks

The evolution of radio access technologies and the user demand for increased capacity drives network deployments towards multiple cell layouts, often referred to as Heterogeneous Networks. With the ongoing rollout of commercial Long Term Evolution (LTE) networks, not only different radio access technologies are offered but LTE networks can also be multi-layered by themselves, consisting of differently sized cells providing coverage in overlapping areas. This comes with increased complexity of network management, which is even more relevant in common multi-vendor deployments, where coordinated configuration and operation of network elements provided by different vendors is essential. In this paper, we investigate and evaluate possible allocation schemes of an LTE radio parameter, the Physical Cell Identity. Results indicate that a particular allocation strategy, the range separation provides an elegant and efficient solution, which makes PCI management in multi-layer, multi-vendor networks easier. The standardization relevance of the range separation scheme is also discussed.

Insight based dynamic QoE management in LTE

Long Term Evolution (LTE) subscribers expect seamless Quality of Experience (QoE) during their ubiquitous Internet access. Due to the high degree of application and content diversity, the amount of system resources that enable good QoE is application session specific and varies on a wide scale. Accordingly, good QoE requires dynamic and adaptive application session level resource management. The standard LTE Quality of Service (QoS) architecture was designed to provide bearer level differentiation thus it has neither the insight to the application sessions nor the granularity to provide proper QoE for each application session. Existing proprietary enhancements operating via context agnostic and static rule based bearer prioritization or intra-bearer scheduling are not able to allocate resources according to the actual QoE requirements of the application sessions. This paper describes a solution that maximizes the level of QoE and uses the system resources efficiently by dynamically adapting the QoS configuration to the estimated demand of individual application sessions. Simulations show that the solution is superior to the existing static methods that fail to provide proper QoE.

LTE relay node self-configuration

Relays are mobile network base stations, which connect to the network via an in-band wireless backhaul link instead of using a dedicated wired or microwave backhaul link as regular base stations do. Connecting through a wireless backhaul raises difficulties when it comes to accessing the operators Operation, Administration and Maintenance (OAM) System, since prior to the appearance of relay nodes, wireless access was reserved solely for user equipments and not used by network elements at all. However, after deploying a relay node, establishing an initial OAM access is essential in any kind of configuration, particularly when considering an automated configuration process known as self-configuration in principle. Self-configuration is gaining more importance for regular base stations as well as relays as automation of OAM processes is seen as a major contributor to reduce complexity and cost in network operation. In this paper, we introduce a conceptual separation of the initial configuration phase and the operational phase and present a detailed concept for automatic connectivity establishment to the OAM system. The transition from configuration to operational phase is also covered. The results have been verified by an event-driven packet based simulator and the proposed method has been accepted by the 3rd Generation Partnership Project (3GPP) as the baseline solution for configuring relays in the next generation radio access technology, Long Term Evolution (LTE).

Efficient LTE PDCP buffer management

Long Term Evolution (LTE) evolved Node Bs (eNB) have dedicated buffers for each Radio Access Bearer (RAB) implemented at the Packet Data Convergence Protocol (PDCP) layer to receive and prepare user data for transmission over the air interface. These buffers multiplex the connections established by the active applications of a single user. Modern applications use the Transmission Control Protocol (TCP) that has a greedy mechanism to transfer data by trying to use all the available resources. If not managed, TCP sources may send large bursts that may eventually overload the PDCP buffer causing tail drops and quality degradation. In order to prevent these incidents, buffer management mechanisms are required. As this paper shows, neither the standard 3GPP Discard Timer nor the conventional Active Queue Management (AQM) techniques are able to properly manage the PDCP buffer. As a solution, two alternative mechanisms are proposed, which are more suited to the PDCP specific environment. Their superior performance is shown with detailed simulations.

Two-way TCP performance issues and solutions in asymmetric LTE radio access

Due to the joint evolution of mobile applications and cloud services, mobile users routinely act as both content consumers and producers. Nowadays there is an increasing trend in both the downlink (DL) and uplink (UL) data traffic, which are predominantly Transmission Control Protocol (TCP) based. Simultaneous DL and UL TCP transmission over an asymmetric link causes significant throughput degradation in the direction with higher capacity whenever TCP data and acknowledgment (ACK) segments are not treated separately. The radio access links of the Long Term Evolution (LTE) systems are asymmetric with higher capacity in the DL than in UL, making LTE prone to the asymmetry problem. In this context, the degradation is caused by multiplexing data and ACKs in the same User Equipment (UE) buffer before UL transmission. This paper proposes and evaluates three novel alternatives that solve the problem at the network side, i.e., transparently to the end devices. Simulation results indicate that the solutions can efficiently detect and resolve the asymmetry problem and they are able to reach the theoretical optimum within 1-5%. Excellent performance is provided even in the presence of complex web traffic.

Integrated Mobility Load Balancing and Traffic Steering mechanism in LTE

Mobility Load Balancing (MLB) is a Self-Organising Networks (SON) use case with the scope of detecting and resolving radio overload. In case of overload, the MLB triggers the handover of cell edge users towards less loaded neighbour cells to better align the traffic demand with the capacity available on the air interface. This, however, also increases the load on the transport links of these cells; therefore, MLB should consider the transport status in order not to cause transport congestion while resolving air interface overload. This paper proposes a general MLB framework that, unlike existing mechanisms, has efficient means to properly consider the transport load and congestion in addition to the radio load. The solution is not limited to any particular transport network topology and it also acts as a Traffic Steering (TS) mechanism as it resolves transport overload by redirecting users to neighbour cells with spare transport resources. The performance of the proposed framework was evaluated with simulations. Results indicate that the solution improves the overall system performance by balancing both radio and transport load.