Optimizing operations remains a key challenge for telecom operators. Ovum's "2016/17 ICT Enterprise Insights" survey of nearly 480 telecom operator executives around the world shows that the need to improve operational efficiency and increase revenue is almost as important.
Telecom operators need to achieve more efficient operations in marketing, sales, customer service, billing and network operations. To support these efforts, telecom operators are considering AI technology for higher levels of automation, optimizing business processes to better serve customers, and optimizing network and traffic management.
Traditional machine learning has fallen behind
AI is an area with a long history, its functions are constantly evolving, and it has recently undergone major breakthroughs. Today, AI systems can perform tasks that are equivalent to or superior to technologies related to human intelligence (such as image and speech recognition, decision making, and language translation). These functions can be implemented in several techniques in machine learning (such as deep learning) and in areas such as natural language processing.
Although telecom operators such as AT&T and Verizon are already using AI/machine learning technology, vendors and telecom operators have not fully realized the most influential advantages of this technology. For example, AT&T has been using machine learning for network management and call center automation. In network management, machine learning systems capture data, analyze data, identify anomalies, and generate tickets for engineers or other service personnel to help them solve problems. However, early adoption is not necessarily that complicated.
In traditional machine learning methods, the programmer needs to specify an event so that the machine learning algorithm recognizes it to make a decision (this process is called feature extraction). This approach limits the accuracy of operations to the experience of the programmer or network team. Challenges such as unknown network conditions that cause network failures will be ignored, and the resulting suboptimal network conditions can seriously affect the customer experience. Telecom operators who invest heavily in SDN (such as AT&T) will need effective AI technology to predict known and unknown system failures and take action immediately before detected failures affect customer service quality. The implementation of this feature requires AI technology that can function without human intervention (reducing the risk of decision making and "man-made" errors).
DRL allows AI to further develop
Deep learning is expected to circumvent the challenges of traditional machine learning methods and represents the cutting-edge technology of today's machine learning. The deep learning system uses a computational model that mimics the working mode of the human brain, so it can perform autonomous learning without the need for a human "teacher" to extract the expected features, which is the network characteristics that the operator refers to.
DRL technology combines deep learning with another type of machine learning called reinforcement learning (RL). RL is a learning that enables machines and software agents to automatically determine the ideal behavior in a particular situation to maximize performance.
Huawei and other vendors will use DRL in network management as one of the important use cases of this technology. Due to the emergence of the Internet of Things and other new digital service products, the number of connected terminals has increased, and the traffic generated by services such as video has increased. Ovum expects that in the five years from 2015 to 2020, the data traffic of telecom operators' mobile networks and fixed broadband networks will grow at a compound annual growth rate of 25%. This growth rate means that the overall market will grow by a factor of three. In terms of mobile networks, this development will be even more extreme, as total traffic is expected to grow by more than seven times.
As telecom operators invest in SDN, the autonomous learning capabilities of SDN controllers will become critical. Since more of the telecommunications carrier network is under the control of the SDN controller, it needs analysis capabilities to support it to adapt to any network scenario and still maintain high network performance. SDN controllers will need to be forward-looking in their operations; make decisions in real time based on current and historical data. Once a decision is made, an impact assessment is required to ensure that the quality of service is not compromised. In addition, when making decisions about network status, these decisions should not be limited to known network failures, but also to unknown network failures. This approach will ensure that the network is optimally operated under the control of the SDN controller.
Lithium Polymer Battery,Charger For Lithium Ion Battery,Lithium Ion Battery 24V,Polymer Lithium BatteryPolymer lithium battery
Langrui Energy (Shenzhen) Co.,Ltd , https://www.langruibattery.com