 We have been appreciating self-organizing networks in the sense that they adapt very quickly under certain policy directives and given KPIs and parameters to network dynamics and channel behavior, which evolves over time. However, with the emergence of machine learning and artificial intelligence, it is important to look at newer technologies, which are complementary rather are altogether new form of self-organizing networks. I am referring to the cognitive radio networks, so we are going to convince ourselves on why this move is important from SONs to CRNs and we look at some interesting definitions and we look at the keyword cognitive, how it is executed as a loop. In self-organizing networks, we know that there are high level parameters and policies to automate the telecommunication and networking infrastructure, but there is a requirement to avail newer emerging technologies to reduce the otherwise minimalist human intervention. It is only possible once a high level goal is given to the network and the network understands those goals and adapts accordingly. This will involve some sophisticated technologies such as artificial intelligence, machine learning, data mining, etc. On the left hand side, we see that it is the network management with closed loop automation. If you recall, when we started motivating ourselves for self-organizing networks, we started off with something called the network operations that would give rise to network workflows, network planning and this resulted into the coordination function between the underlying infrastructure with performance management, configuration management and fault management. And then we saw that human effort actually was more involved in the planning and policy and the system was automatically reporting the statistics and the network behavior was getting automated. So human involvement was only there to declare policies and give out the KPIs. But there is a caveat. The caveat is the network is not learning much from the experiences that it has acquired over time. So the requirement is to look at the goals which the humans define and pass these goals to the network. The network uses an intelligent cognitive process involving machine learning, artificial intelligence, data mining, etc. And then measures the network performance and adapts the parameters accordingly to improve the user experience at the same time planning in the light of the experience that it has to improve its behavior further. This cognitive approach is known as the cognitive network management loop. We are going to look more into it in due course of time. Let's look at the word cognitive. Action is the ability to think. We all humans are cognitive by design. The cognitive radio was the classical form of radio spectrum sharing and access mechanism which was predominantly related to the physical and MAC layer. So what was happening was using some kind of sensing, actuation was taking place at these two layers. Then came the concept of cognitive network. Cognitive network is a very broad concept that relates to the perception of the network conditions and then planning and deciding and actuating correspondingly. So cognitive networks is a very broad term. The scope is covering literally all aspects including the physical aspect, the wireless aspect and the core services as well. So here the network learns from adaptations to make future decisions and the overall focus is to reach to the human defined or the manager defined goals. The cognitive radio network which is the focus of our lecture today is similar to cognitive networks. The underlying restriction or the constraint that we draw is that it is mostly related to wireless network. We are going to look at the cognitive process that takes place in cognitive radio networks and how learning capabilities can be introduced within these networks. This diagram is a beautiful illustration along two axis. On the X axis we have the scope. The scope is in terms of functionality. So we know self-organizing networks are predominantly related to the wireless part. Cognitive radios, cognitive radio networks are what is our definition. Cognitive network is a very broad term. So it encompasses literally the cognitive radio network and the cognitive radio technology as well. On the Y axis we have the level of cognition or the level of perception dependent decision making based on past experiences. So we see that sounds were at the very primitive level but as soon as or the moment we describe cognition the level of cognitiveity increases. The cognitive loop can be thought of as some keywords. Those keywords are the environment in which the channel and the network dynamics are evolving over time. Sense as a keyword, learn, act, plan, decide. And goals is what the human or the system administrator provide. So sense is at the heart. The cognitive process continuously monitors the environment. And then this particular information is used to create strategies. So we see that plan is getting input from sense. On the basis of it the system, the overall cognitive radio network builds up knowledge on the effects of the actions which were taken place so it learns from past experiences. In the light of those experiences it decides what plan should be put into action and then invokes or activates the plan using the APIs that is to actuate. So this cognitive loop helps to move in a very big stride from the self-organizing networks to cognitive radio network. The book by Seppo Himalayanan is providing a beautiful discussion and also cites references on cognitive radio network. You might like to as well have a look.