Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Machine Learning for Cognitive Radio

No description
by

Afaque Hussain

on 10 November 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Machine Learning for Cognitive Radio

Machine Learning For Cognitive Radio
Afaque Hussain
Introduction
• "Spectrum Drought"

• Solution?

• A radio that can adapt and configure itself dynamically.

• Can a normal radio suffice?
Non-Cognitive Systems
• All possible actions programmed.

• Good at what being programmed for.

• Given these two statements,
1. If today is Monday, then you have to go to work.
2. (Unfortunately) Today is Monday.

• Can these systems infer an action form these?
Need for Cognition
• Dynamic Environments and Radios

• Software Defined Radios (SDR)

• Cognition = Awareness + Reasoning + Learning + Adapting

• SDR + Cognition = Cognitive Radio
Abstract Architecture of a Cognitive Radio
P →-> Q
P
Q
Artificial Neural Networks
• Inspired by the Nervous System.

• Artificial Neurons.

• Dynamic Networks among neurons.
Artificial Neuron
• One or more weighed inputs.

• Activation Function.

• Single output.
Network & Training
• Weighed Connections

• Training

• Algorithms alter the weights among connections.
Application to Cognitive Radio
• Performance Characterization

• Spectrum Sensing

• Spectrum Classification
Parameter Optimization Problem
Meta-Heuristic Algorithms
Genetic Algorithms
Simulated Annealing
Tabu Search
Ant Colony Optimization
Application to CR
Hidden Markov Models
Logic Deduction Systems
Application to CR
What if the Cognitive Radios fall into infinite Adaption due interaction with other Cognitive Radios?
Game theory
Application to CR
Challenges
Conclusion
• Genetic Algorithms

• Simulated Annealing

• Tabu Search

• Ant Colony Optimization
• Search Heuristic

• Random Candidate Solutions

• Fitness Function

• Generations by Mutation

• Several Generations produce optimal solution.
• Inspired by the Annealing process in metallurgy.

• The probability of accepting worse solution is kept high, initially.

• Probability is slowly decreased to attain optimal solution.
• Improves Local Search Algorithms.

• List called 'Tabu' (forbidden).

• Avoids Local search algorithms getting stuck in sub-optimal solutions.
• Inspired by the behavior of ants.

• Pheromone Trails

• Follow the stronger trails

• Finds the optimal path over time.
• Dynamic Environment

• Quality of Service Requirement changes with time.

• Performance Optimization

• Convergence time increases with increase in complexity.
• Models a system whose state is hidden and only outputs are visible.

• Urn Problem

• Used for Modeling complex Random Phenomenon.

• Used to Model Primary User usage of the channel.
• Rule Based Systems

• Ontology Based Systems

• Case Based Systems
• Used to Logically deduce new rules.

• Updates the Knowledge base.

• Experience of CBS improves with the passage of time.
Application of Various Algorithms
• Models for conflict and cooperation between intelligent agents.

• Predicts the behavior of such intelligent agents.

• Set of players.

• Set of Moves.

• Payoffs for each combination of moves.
• Prisoner's dilemma game.

• Model Interaction of various Cognitive Engines.

• Prevent Infinite Adaptations.

• Prevent Sub-optimal performance.
• Field still in Infancy.

• Basics and experimental Implementation.

• Complexity and Performance trade-off.

• Prevention of Infinite Adaptations.

• Fault tolerance.

• Security.
• Need for Cognition in a Radio.

• Various Machine Learning Techniques.

• Game theory.

• Appropriate Artificial Intelligence must be used for particular application.

• Implementing a Mitola Radio is quite complex and difficult.
Overview
• Introduction
• Artificial Neural Networks
• Meta-Heuristic Algorithms
• Hidden Markov Models
• Logic Deduction Systems
• Game Theory
• Challenges
• Conclusion
Dynamic Environments
ANN, GA, HMM
RBS, OBS, CBS
RBS, OBS, CBS
Full transcript