Publication Type : Conference Paper
Publisher : IEEE
Source : 2026 International Conference on Smart Futuristic Technology
Url : https://doi.org/10.1109/icsft66733.2026.11507660
Campus : Amritapuri
School : School of Engineering
Center : Humanitarian Technology (HuT) Labs
Department : Electronics and Communication
Year : 2026
Abstract : Edge devices are becoming more burdened with the task of executing AI inference in real-time, although they have limited energy and efficient computing capabilities. The memory access and data movement is a very important part of the overall energy consumption and effective memory design is an important factor in development of low-power, high-performance AI applications. This Paper offers the design and simulation of an energy-efficient hybrid cache memory system based on SRAM and ReRAM energy-efficient edge-based AI inference. The hybrid architecture employs the high-speed access speed of SRAM to the data required to be accessed frequently, and the non-volatility and low leakage power of ReRAM to less frequently used data. The entire system was coded in Verilog HDL and several verification of functionality and performance was conducted in ModelSim at a hardware level. It incorporates SRAM and ReRAM cache banks in the architecture, an LRU replacement unit, and frequency-based data placement mechanism to provide energy and access latency optimization. The results of the simulation are proven to be the correct work of a cache with the overall hit rate of 63.33% and the decreased energy use in comparison with the DRAM-only case. The experimental results of an analytical estimation of energy consumption show that in 60 accesses the hybrid cache uses a total of 223.8 pJ of energy, which shows that the hybrid cache has the potential of being used in a low-power high-performance edge computing system
Cite this Research Publication : Navaneeth Krishnan, Ram Krishna, Rajesh Kannan Megalingam, Energy-Efficient Hybrid Cache Memory Design for Edge-Based AI Inference Using SRAM and ReRAM, 2026 International Conference on Smart Futuristic Technology, IEEE, 2026, https://doi.org/10.1109/icsft66733.2026.11507660