A New Cross Ring Neural Network: Dynamic Investigations and Application to WBAN

Wireless body area network (WBAN) is a crucial tool in modern medical areas. This refers to the intelligent interconnection of wireless sensor nodes installed outside or inside the human body to recover some human vital signs. The security and bandwidth saturation remain crucial problems of this technology. In this work, a cross-ring-based neural network model (CRNN) is obtained from the Hopfield neural network definition and its complex dynamics are deeply analyzed. The steady-state study shows that the model has a unique equilibrium point and its analysis shows that the model’s dynamics is self-excited. Based on the two-parameter Lyapunov spectrum, the set of synaptic weights has been quickly identified to illustrate hyperchaotic behavior of the CRNN. With the help of bifurcation plots, graphs of the Lyapunov spectrum, and phase portraits, we have characterized periodic, chaotic, and hyperchaotic patterns in the model. Furthermore, an experimental setup has been built using microcontroller technology to further support the results of the numerical simulations. A parallel compressive sensing algorithm combined with a nonlinear congruent generator has been used to show the application of the newly designed CRNN to medical image compression and security. Security and compression performances indicate an efficient scheme with the capability to resist various attacks and the capability to produce low-size data image from large-size data image. For instant from
$256\times 256 $
image size, an encrypted and compressed output image is obtained at the compression rate of 0.5, processing time of 0.147 ms, encryption throughput of 1783.3 MBits/s, for a 2.9-GHz processor, the number of cycles to process the algorithm is 1.62. Consequently, the proposal can be applied to WBANs.