Autonomous Defensive Space Control via On-Board Artificial Neural Networks
Abstract
Future advances in neural network technology, coupled with increased computer processor capability, may create an opportunity to develop systems that enable satellites to autonomously differentiate, detect and defend against attacks. The Air Force should take advantage of this potential opportunity by investing the necessary resources for the development of space-based neural networks. An artificial neural network (ANN) or commonly just neural network (NN) is an artificial intelligence system created to mimic the ways and methods in which our own brains respond to and learn from inputted stimuli. Each of these networks consists of an array of neuron-like gates programmed to take action once a designated threshold is crossed. These ANN are adaptive, and learn through continued processing of inputted stimulus while developing a memory by storing the actions it takes in response to this stimulus. This memory gained through storing data enables ANNs to become somewhat autonomous over time because they have the ability to recall a given action taken based on a given input received. Computer processing will likely continue to increase in power while decreasing in size. Expanded processing capability could potentially enable the placement of neural networks, requiring significant processing power and storage capacity, on-board satellites that must contend with size and weight limitations. At the same time, advances in the fidelity and sensitivity of neural network capabilities might give spacecraft processing units (spacecraft brain) more "intelligence," or ability to give raw data meaning. The merging of increased processing power with a reliable neural network will potentially give a spacecraft the ability to recognize, through its telemetry, that something is attacking it.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 2007
- Accession Number
- ADA497501
Entities
People
- Michael T. Manor
Organizations
- Air War College