Artificial Intelligence has come to an age enabling new possibilities in human convenience. Voice assistants, language translation, speech recognition, computer vision, handwriting recognition are examples where AI has exhibited significant progress. As the ecosystem evolves and new algorithms on statistical learning, blended with domain knowledge takes shape, there emerges new possibility on “autonomicity”. A system is said to be autonomic if it can manage its own affair by making decisions based on the situation. Autonomicity is defined as the property of being autonomic or autonomous. Cloud and Virtualization bundled with big data technologies are enabling new ways of designing and deploying network for diverse use cases like medical services, critical machine operations in industry 4.0 or ultra-high bandwidth consuming applications. Network connectivity has become the new oxygen to the cloud based connected entities. There are challenges in enabling different types of network connections with a fixed control logic which limits the speed of adoption. Here 5G is revolutionizing connectivity paradigm with the concept of network slice that enables composing one based on the use case that the user demands. This user-centric networking will have dynamic variations and a fixed control and management is likely to fall short and cognitive network control and management takes prominence.
Learning and reasoning capabilities drive cognition. The reasoning is synonymous with analysis of information based on a knowledge model to get insight from the information. Reasoning enables decision making to meet the goal. For real-time decision making, real-time information analysis plays a critical role.
As AI and ML come to aid in automating the network management and make networks autonomic, similar to an autonomous car, there are challenges ahead in taking full advantage of the technology advances in machine learning. Machine learning algorithms are proven to work in automating well-defined tasks like playing a game of chess, learning patterns for a large set of voice samples or a large set of images. If there are intricate domain knowledge bundled in multiple layers, the semantic understanding of the domain becomes critical where the domain ontology plays a role. Ontology helps represent the domain knowledge. If the complex concepts are embedded into the ontological representation, it gives the autonomic system the power to reason more effectively and intuitively as a learned human does. Blending domain knowledge with statistical learning is a fast progressing research area. This is an interesting yet complex research topic.
Cognition in networking domain raises a lot of interesting questions. Will cognition give rise to BOTs to perform the role of a “Network Architect”, “Network implementer”, or a “product designer”? Would human architects and implementer come under the challenge of artificial intelligence? Write to me with what you think. Looking forward to hear from you on your thoughts.
