Generation Mechanism and Grid Modulation of Multi-Structural Attractors in a Locally Active Memristive Map Neuron


Xu X., Liu L., ERKAN U., TOKTAŞ A., Cao Y.

IEEE Internet of Things Journal, 2026 (SCI-Expanded, Scopus)

  • Yayın Türü: Makale / Tam Makale
  • Basım Tarihi: 2026
  • Doi Numarası: 10.1109/jiot.2026.3744195
  • Dergi Adı: IEEE Internet of Things Journal
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, ABI/INFORM, Aerospace Database, Compendex, INSPEC, Technology Collection (ProQuest)
  • Anahtar Kelimeler: DSP implementation, Image encryption, Memristor, Multi-structural attractor, Tunable multistability
  • Ankara Üniversitesi Adresli: Evet

Özet

Multi-structural attractors have received considerable attention in continuous neural networks, but systematic research on their generation mechanisms and regulation methods in discrete neural networks is still lacking. To address this issue, this paper proposes a discrete memristive neural network (LCY-M) model, which introduces a locally active non-polynomial memristor (LANPM) into a two-dimensional map neuron framework, thereby achieving the generation and regulation of extendable multi-structural attractors. Based on the fixed point analysis, this paper reveals the formation mechanism of multi-structural attractors in the model, and introduces multi-level logic pulses according to the unidirectional migration property of the fixed points to achieve the modulation of planar grid-extended multi-structural attractors. The research results demonstrate that adjusting the relevant parameters can effectively change the positions of the fixed points in phase space, thereby inducing complex dynamical phenomena such as unidirectionally extended multi-structural attractors, grid-extended multi-structural attractors, spatial offsets, and extreme multistability. Subsequently, this paper verifies the feasibility of the digital circuit implementation of the model through a Digital Signal Processor (DSP) platform. Finally, taking advantage of the rich multistability and complex attractor structures of the LCY-M model, a color image encryption algorithm was designed. The average information entropy of the three independent color channels in the ciphertext image reached 7.9993, indicating that the model possesses excellent image encryption performance.