Use cases and challenges in telecom big data analytics

Abstract

This paper examines the driving forces of big data analytics in the telecom domain and the benefits it offers. We provide example use cases of big data analytics and the associated challenges, with the hope to inspire new research ideas that can eventually benefit the practice of the telecommunication industry.Keywords: Big data, Analytics, Telecom, QoE I . I N T R O D U C T I O NThere has been much hype about big data analytics -a collection of technologies, including the Hadoop distributed file system, NoSQL databases, and machine learning tools. One study estimated that it can generate hundreds of billions of dollars of value across industries [1]. Another study reported that 75 of telecom operators surveyed would implement big data initiatives by 2017 [2]. Every operator is seeking new ways to increase operational efficiency and marketing effectiveness by leveraging big data technologies. But the question is how does big data analytics differ from prior arts such as data warehousing and statistical methods, in terms of the capabilities of uncovering insights from large volume of datasets? What are the compelling use cases in telecom, and what are the challenges?This paper provides a retrospect on how telecom operators have been striving, before the era of big data, to analyze large volumes of data in order to support their business and operation. We then examine the driving forces of big data analytics in the telecom domain and the benefits it offers. Finally we provide example use cases of big data analytics and the associated challenges, with the hope to inspire new research ideas that can eventually benefit the practice of the telecommunication industry. Corresponding author: C. Chen Email: cchen@iconectiv.com processing (OLAP), and data mining are adopted by telecom carriers to improve operation efficiency and user experience. To appreciate that, it helps to understand how a telecom network is managed. Figure 1 shows a simplified telecom management framework adopted from the TM Forum [3]. The framework contains three horizontal layers -resource, service, and customer, spanning across two vertical perspectives -infrastructure & product and operations. Examples of big data analytics use cases are shown in places according to their nature.The resource layer includes activities related to network build-out, planning, and monitoring. Operators constantly monitor performance of the networks (including user devices and network devices such as routers, switches, base stations, etc.) in order to assure smooth operation. Data collected at this layer includes alarms generated by the network devices and key performance indicators (KPIs) such as packet loss ratio, latency, traffic load, etc. The datasets support use cases for network planning, capacity management, and fault management.The service layer includes activities related to provisioning of user services (voice, data, and video). It also supports proactive monitoring and reactive diagnostics required by service-level agreements -a contractual agreement betw...

Identifying Similarities of Big Data Projects–A Use Case Driven Approach

“…The next cluster, derived from the hierarchical clustering results, consists of three cases, namely no. 15 [52], 28 [74] and 29 [46]. The general aim of those cases can be described as the integration of data from different sources, improving the scalability and leading to better analysis results.…”

“…In the past, data mining methods were used in the telecom area to figure out problems only in an isolated way, for example fraud detection based on call detail records. In order to consider different telecommunication aspects, nowadays various information, coming from mobile networks, GPS devices and social media has to be considered [74]. The target of the last case study in this category is to turn regular factories into smart factories, where resources and machines communicate and deliver smart products that are aware of their production history.…”

“…The data used in this cluster can be both, structured and unstructured. Examples for structured data can be records of the log and machine operating times [46,74]. In contrast, social media data, camera pictures, and sensor data are examples of unstructured data [46,74].…”

“…Examples for structured data can be records of the log and machine operating times [46,74]. In contrast, social media data, camera pictures, and sensor data are examples of unstructured data [46,74]. Furthermore, all three case studies maintain personally identifiable information, which can require special processing techniques.…”

“…Furthermore, all three case studies maintain personally identifiable information, which can require special processing techniques. This information can be the location of a driver, smart card data of public transport users, or the call history log of a telecom company customer [52,74]. Permanent as well as transient data (data that is deleted by the end of a session) play an important role in the performed analysis in this cluster.…”