We propose to develop the Sensing portion of a Self Organizing Resilient Network (SORN-S) by integrating existing natural-system process knowledge with a new affordable, massively-parallel, pattern-detection technology. The biological immune system (BIS) has been offered as a model by many who see its natural fit to cyber networks viewed as organisms, but computational performance limitations have obstructed necessary high fidelity translation. The cortical feed-forward/backward hierarchy is understood as a powerful pattern learner/predictor process, but computational translations remain complex and expensive. Both are compatible multi-agent architectural models that can be combined and distributed among existing network endpoints, potentially embedded in new systems designs, or packaged as a USB device for legacy add-on. A high fidelity BIS model adapted to cyber-networks offers both signature and behavior based detection of packet-born attacks and endpoint infections, and notably features the ability to detect previously unseen attacks and infections.The proposed project will explore and evaluate the integration of three levels of nested pattern learning/detection architectures enabled by the new pattern-processor technology; and show feasibility for rendering the result as an end point add-on device for legacy environments, and potential for integrated inclusion in new network/endpoint design.
