Artificial Intelligence in Telecom Networks
|出版日期||內容資訊||英文 24 Pages, 3 Tables, 1 Chart, 11 Figures
|通訊網路的人工智能 Artificial Intelligence in Telecom Networks|
|出版日期: 2017年10月19日||內容資訊: 英文 24 Pages, 3 Tables, 1 Chart, 11 Figures||
Artificial Intelligence and Machine Learning are the new hot topic in telco networks and for a good reason: They are sitting on a massive pyramid of diverse customer and network data where Artificial Intelligence (AI) can be used to understand, optimize, and improve business and network capabilities. This research report explores the intersection of different AI technologies with telco operations. It examines use cases within Customer Management, Cybersecurity, and Network Automation wherein AI can prove to be very effective to telcos. It constructs an AI implementation roadmap for telcos in their journey to be "next-gen" service enablers and covers a few leading vendors that are providing AI technology to telcos worldwide.
The concept of artificial intelligence (AI) was founded in 1956 as an academic discipline, through a confluence of ideas in neurology, cybernetics, information theory, and computational theory. AI is a new data science aiming to mimic human intelligence processes to some degree, i.e., a human's ability to sense, learn, reason, and take action in real time. It harnesses vast amounts of information gathering and computational abilities to extract and identify hidden data patterns more efficiently than humans. The goal is to detect and diagnose problems, situations, and hazards in their embryonic stage of development, and then make appropriate predictions and prescriptions with greater accuracy. AI use cases have rapidly expanded from traditional robotics to consumer applications, education, finance, healthcare, law, manufacturing, and many others, but most importantly and most recently, telcos.
Two AI approaches, in particular, have gained footprints into telco applications, namely machine learning (ML) and natural language processing (NLP). ML is a data science that enables a computer to autonomously learn and make predictions on specific data inputs and patterns without human intervention. NLP is a computer science that enables machines to understand and perceive human languages and expressions (e.g., text, voice, visual, and sentiments). Deep learning (DL), on the other, hand is a subset of ML, which focuses on data representations and optimization based on some degree of supervised learning. The learning somehow depends on the way the data information is interpreted and judged using some symbiotic assumptions. DL applies to the environment with large- scale data. The learning is done via efficient reorganizing of the data representations, a process commonly called data training. Training accuracy depends of the amount of data used. DL is a very popular approach for a number of applications, including voice and audio recognition, computer vision, pattern and object recognition, and machine translation. All of these can create immense value for telcos, and automate tasks that required time-consuming human involvement. Moreover, ML, NLP, and DL can create new value, such as processing the vast amounts of data that telcos collect.
The primary criterion for optimally using AI technology to its full potential is massive amounts of data. Data can be in any form: structured, unstructured, or semi-structured. The heterogeneity of data is typically characterized by the "3Vs" (often used to define Big Data), i.e., extreme volume of data, wide variety of data, and sheer velocity of processing data. This is precisely why and where telcos hold a unique advantage. Telcos satisfy all 3Vs, which is fundamental to any AI technology. By leveraging the massive pyramid of diversified data accumulated across business, customer, and network levels, both in real time and over a period of time, telcos can use AI technologies to automate business