Smart network planning for MBB TextStart Fast-growing mobile data services and smart devices have brought operators a large number of new subscribers and a significant increase in network traffic, but the revenue has not grown proportionately. And when they look to new business models as a way out, they are often plagued by the signaling storms and unpredictable data traffic changes caused by smart devices and the applications running on them. The root causes Industry studies and estimates suggest that, in the future, traffic from a smart device may be 10 to 100 times higher than that from a feature phone. This requires operators to make enough network investments to meet ever-increasing network capacity requirements. From a global perspective, even leading operators have suffered unexpected network overloading to different extents over the past couple of years, even after spending heavily on their networks. The overloading can occur at the wireless side or the core side. Nonetheless, it is impossible for operators to increase network investments year after year without an end. The traditional method for network planning, based on service traffic growth only, has failed to solve this industry dilemma for the following reasons: Before smart devices appeared, devices behaved in a relatively unified way, as services running on them were voice services and simple data services. As a result, the traffic model on the network equipment side was relatively stable during a given period of time. When planning networks by predicting network development trends, operators had only to focus on the volume of transmitted data based on the number of subscribers at all sides other than at the wireless side, where they had to consider coverage and interferences. Nonetheless, smart devices have brought numerous, unpredictable changes, such as signaling width, the maximum numbers of concurrent connections of numerous new services, changes in the data throughput of every data service connection, and the background traffic of new services. As previous networks had only one variable – the subscriber count, network planning was simple and could be easily done by using this function: F (future demand for network capacity) = current network capacity + the increment of network capacity (in proportion to subscriber growth). Today, smart devices bring with them diverse behaviors and services. Therefore, a method is needed that can recognize changes in the subscriber count, devices, and services and that can predict network resources development for network planning: F (future demand for network capacity) = current network capacity + the increase in network capacity (in proportion to subscriber growth) + the increase in network capacity (smart devices/service growth). It is obvious that current network planning should be based on the projection of changes in the traffic model brought by the subscriber count, devices and services. However, how to visualize the impact of smart devices and services on the capacity remains a problem. First of all, the visualization method must be able to predict dynamic changes in the traffic model brought by smart devices and services – changes in the equipment’s traffic carrying capacity. This capacity depends on end-to-end network equipment resources, whose theoretical limit is the scenario where different types of traffic are most effectively reused on the equipment’s software and hardware resources. The traffic model consists of parameters that represent statistics of the performance and times of use of different resources on network equipment and that indicate the actual utilization of the existing equipment’s specified capacity. Smart devices have brought about many parametric changes that lead to dynamic changes in the traffic model for network equipment and, consequently, in the equipment’s ability to support resources. Any failure to find resource bottlenecks in the changing network in a timely manner may lead to unexpected network overloading. Secondly, the method must be able to match changes in the traffic model to the end-to-end capacity of network equipment, recognize weaknesses, and perform precision network planning, thus answering the two critical questions that operators have to face when predicting long-term network development: When should timely investment be made? How can network capacity expansion be planned properly? Practice In December 2010, operator D in country G began working with Huawei to design a solution addressing those challenges. It was called the MBB capacity service solution, comprising two major parts. Prediction of the traffic model on the network: This part correlated service development with traffic model development and predicted quantitative changes in the traffic model in the following two steps. Visualizing the impact of smart devices and services on the network: Based on the analysis of its historical network data, the project team studied the impact of the addition to operator D’s wireless nodes B, the enablement of the CELL_PCH feature and the use of smartphones on the traffic model (traffic model) for the equipment. Predicting the traffic model for PS equipment of the core network: The project team predicted the traffic model for its core network equipment in 2011 based on operator D’s business development plan for the year and in combination with an analysis of its existing networks and a database about the relationship between smart devices and the traffic model that had been created using Huawei’s global experience. Test of network resource capacity: This part correlated the traffic model with the capacity of equipment resources and provided recommendations for network adjustments needed to satisfy future capacity requirements, which also involves two steps. The first step was to analyze future equipment capacity requirements. Operator D faced complex changes in its actual network traffic. As networks are gradually going IP, there is no precise, fixed algorithm that can translate the traffic model into equipment resource utilization. The project team tested every piece of SGSN equipment in the lab environment, ascertaining future equipment capacity requirements – a best industry practice for increasingly accurate calculations of the traffic model. The second step was to identify the requirements for equipment resource capacity. The project team predicted correlations between equipment capacity and resources through the traffic model before formulating operator D’s development plan for 2011, which answered the questions of whether its existing network platform could support its business development and when it would encounter capacity bottlenecks. As a result, the operator was able to know in which category of network resources overloading would take place, when would be the best time for network adjustment, how network capacity expansion could be planned properly, and when a timely investment should be made. By analyzing root causes and effects of actual changes of smart devices and services in mobile data networks through the above steps, operator D successfully predicted the traffic model for its networks and equipment and developed plans to address the future use of equipment capacity, while minimizing the risk of abnormal network overloading. Outlook With the ongoing development of smart devices and services, operators have come to realize that their plans for future services need to allow for the impact of these services. Our experience from this operator D project suggests that operators need the following critical capabilities in this respect: Capability of visualizing network-wide devices and services. The ability to accurately understand the behavior of mainstream devices and services in a real network environment is the basis for assessment in daily monitoring. Therefore operators particularly need the ability to visualize the changes in the traffic model before and after network adjustments as well as different effects of devices and services on such changes. Capability of business development planning and analysis based on devices and services. When planning services, operators need to think about device and service plans in addition to traditional subscriber growth. Ideally, they are able to break their business development plans down into device and service development in every major area in combination with daily device and service visualization. Capability of designing network traffic control policies. In network investment and construction, equipment resources must be maximized whenever possible before the overall profitable business model for mobile data services becomes clear. In addition to conventional capacity expansion design and differentiated market strategy development, operators need to consider the QoS plan and design. Also, they need to take into account the application scenarios for service QoS use and scheduling, and control policies in an end-to-end way so as to realize differentiated scheduling of subscribers, devices, and services in the network. Network planning based on smart devices and the traffic model is one of the top priorities for mainstream operators. As market strategies and actual network environments vary from region to region, implementation schemes require customization. Huawei is willing to partner with operators to explore best practices in this regard. TextEnd