17.7 SUMMARY

In this chapter, we have described GN as a new form of neural network that can operate effectively within resource-constrained network and adapts to the changing network conditions automatically. The GN implements a fully distributed single-cycle associative memory within the network. Hence, it is highly suited as an online system for providing rapid responses to the changing network conditions. It is also highly scalable, hence a distributed IDS based on the GN would assimilate vast amount of information on the network traffic patterns without running out of memory resource or becoming sluggish over time. The simple GN approach has been extended, using the HGN model, into DHGN. This distributed form of HGN would be useful for developing self-aware WSN and MANET for threat detection. We have shown through a case study how the simple GN approach could be used for DDoS attack detection in WSNs. The DHGN provides better framework for network deployment and can handle noisy/distorted patterns. Hence, the concept of network thresholds developed for WSNs may be readily applied within MANETs using the DHGN. The mobility consideration and the sparsity of the nodes in a MANET (in contrast with WSNs) are addressed through the standard or the variable forms of the DHGN.

Get Mobile Intelligence now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.