Reducing Power Consumption in AI Edge Devices: A Low-Power Design Approach

Reducing Power Consumption in AI Edge Devices: A Low-Power Design Approach
Niranjana Gurushankar

The rapid evolution of AI technologies has sparked a transformative wave in various industries, with energy-efficient AI edge devices at the forefront. These devices, designed for localized processing with minimal power consumption, are becoming integral to applications such as IoT, wearables, and autonomous systems.

Behind this innovation lies the expertise of Niranjana Gurushankar, whose work in designing and verifying Application-Specific Integrated Circuits (ASICs) has significantly contributed to advancing low-power solutions for AI edge devices.

As Niranjana puts it, “Energy efficiency is not just a technical challenge; it's a commitment to sustainable innovation that drives meaningful change across industries.”

With a strong foundation in low-power design principles, she tackled the complexities of ASIC development with a focus on reducing power consumption. Through techniques like clock gating, power gating, and voltage scaling, she has enhanced the efficiency of these devices. Her experience spans multiple levels of the design hierarchy, allowing her to identify optimization opportunities and refine energy usage. “Every Nano-watt saved in a device contributes to extended battery life and a greener future,” she shares, emphasizing the importance of her work.

Debugging complex systems is another area where she excelled, honing her analytical skills to resolve intricate issues in ASIC designs. This involves analyzing waveforms, tracing signals, and using advanced debugging tools to ensure reliability and robustness, critical attributes for AI edge devices operating in demanding environments. “Debugging is not just about fixing errors; it's about building resilience into the system,” she explains, highlighting the meticulous effort behind each successful design.

Collaboration has also played a pivotal role in her achievements. Working closely with firmware teams, she has ensured seamless integration between hardware and software, optimizing power consumption at every step. This collaborative approach has enabled the co-optimization of hardware and software, reducing unnecessary power overhead and enhancing device functionality. “Teamwork bridges the gap between vision and execution, especially in a multidisciplinary field like ours,” she remarks.

By writing comprehensive tests for on-chip components, she increased functional coverage from 75% to an impressive 95.3% using advanced tools like Cadence IMC. This meticulous testing framework has enabled early identification and mitigation of power-related bugs, directly translating to reduced power consumption and improved device performance. “Functional coverage is a measure of how well we’ve anticipated real-world scenarios, it’s our assurance of quality,” she notes.

Her expertise extends to FPGA testing and validation, where she designed comprehensive test plans to evaluate power consumption and validate low-power features. These efforts have helped uncover power inefficiencies early in the development cycle, leading to optimized designs. Her work in FPGA environments has also facilitated real-time measurement of power consumption, enabling precise adjustments for greater efficiency. “FPGA testing is like a rehearsal before the final performance, it helps us perfect the design,” she explains.

Implementing test vectors for Automated Test Equipment (ATE) and High-Temperature Operating Life (HTOL) testing has been another critical aspect of her contributions. These rigorous testing methods ensure the reliability and longevity of ASICs, even under extreme conditions. By identifying potential weaknesses and validating design robustness, she has contributed to extending the operational life of AI edge devices while reducing power-related vulnerabilities. She reflects, “Reliability is the cornerstone of trust in any technology, we owe it to the end-users.”

Her commitment to continuous learning has been instrumental in staying ahead in a rapidly evolving field. Proficiency in tools like SystemVerilog, UVM, and Cadence IMC, coupled with a deep understanding of power optimization strategies, has allowed her to address emerging challenges effectively. “In this industry, learning is not optional; it’s the key to staying relevant and innovative,” she affirms.

Despite the technical challenges, she consistently achieved great results, whether by developing robust UVM test benches for thorough verification or by achieving high toggle coverage to ensure design integrity. Her dedication and expertise have not only advanced the field of low-power AI edge devices but have also set benchmarks for energy-efficient design practices.

Through her work, Niranjana Gurushankar showcases the innovative spirit driving the next generation of AI technologies. Her contributions have not only improved the functionality and efficiency of AI edge devices but also underscored the importance of sustainable practices in technology development.

As she aptly puts it, “The future of technology lies in achieving more with less, more performance, more reliability, and less power consumption.”

How Artificial Intelligence Is Transforming Business
Artificial Intelligence is a critical factor in the strategy of those who want to expand their business impact in this digital era to make a win.

WIDGET: questionnaire | CAMPAIGN: Simple Questionnaire 

Must have tools for startups - Recommended by StartupTalky

Read more