Small Language Models and Neuro-Symbolic AI in Zonal Architectures: The Rise of Small Language Models (SLMs) in Constrained Environments
DOI:
https://doi.org/10.63282/3050-9246.IJETCSIT-V7I1P141Keywords:
Small Language Models (Slms), Zonal Architecture, Neuro-Symbolic AI, Software Defined Vehicle (SDV), Edge Computing, Adaptive AUTOSAR, Deterministic Safety, Real-Time Systems, Knowledge Distillation, ISO 26262, Time-Sensitive Networking (TSN)Abstract
The automotive industry faces a pivotal moment with the rise of Software-Defined Vehicles (SDVs), Zonal Electronic/Electrical (E/E) Architectures, and Generative AI. As vehicle architectures shift from centralized domain structures to decentralized zonal networks to handle more complex wiring and data flow, the need for local intelligence grows. However, deploying Large Language Models (LLMs) on resource-limited vehicle edge devices is challenging due to high computational latency, energy demands, and safety concerns. This paper explores the potential of Small Language Models (SLMs)—under 7 billion parameters, such as Microsoft’s Phi-3 and TinyLlama—as a way to embed advanced reasoning directly into Zonal Controllers. It also tackles the core issue of balancing the unpredictability of generative AI with the strict determinism required by safety standards such as ISO 26262. We suggest and evaluate Neuro-Symbolic AI as an essential layer of the architecture, using symbolic logic to verify neural network outputs in real time. By examining hardware options (NXP S32, TI TDA4), virtualization tools (Adaptive AUTOSAR), and safety protocols (RoboGuard, SYNAPSE), this study shows that SLMs, guided by symbolic logic, can support resilient, decentralized vehicle autonomy, reduce dependence on cloud connectivity, and uphold safety integrity.
Downloads
References
[1] Texas Instruments, "How a Zone Architecture Paves the Way to a Fully Software-Defined Vehicle," TI Tech. White Paper, Oct. 2024. [Online]. Available: https://www.ti.com/lit/spry345
[2] NXP Semiconductors, "Zonal Architecture & Software Defined Vehicles," NXP Training Pres., 2024. [Online]. Available:(https://www.nxp.com/docs/en/training-presentation/TP-TD24-EUF-AUT-T4753.pdf)
[3] Ravichandran, Z., Robey, A., Kumar, V., Pappas, G. J., & Hassani, H. (2025). Safety Guardrails for LLM-Enabled Robots. ArXiv. https://arxiv.org/abs/2503.07885
[4] Z. Lu et al., "Small Language Models: Survey, Benchmark, and Case Studies," ACL Anthology, 2025. [Online]. Available: https://aclanthology.org/2025.acl-long.718.pdf
[5] Castagnone, A., & Nitti, G. (Jan 2026). A Neuro-Symbolic Framework for Ensuring Deterministic Reliability in AI-Assisted Structural Engineering: The SYNAPSE Architecture. Buildings, 16(3), 534. https://doi.org/10.3390/buildings16030534
[6] Abdin, M., Aneja, J., Awadalla, H., Awadallah, A., Awan, A. A., Bach, N., Bahree, A., Bakhtiari, A., Bao, J., Behl, H., Benhaim, A., Bilenko, M., Bjorck, J., Bubeck, S., Cai, M., Cai, Q., Chaudhary, V., Chen, D., Chen, D., . . . Zhou, X. (2024). Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone. ArXiv. https://arxiv.org/abs/2404.14219
[7] Zhang, P., Zeng, G., Wang, T., & Lu, W. (2024). TinyLlama: An Open-Source Small Language Model. ArXiv. https://arxiv.org/abs/2401.02385
[8] Texas Instruments, "TSN in Automotive Zone Architectures," TI White Paper, Oct. 2025. [Online]. Available: https://www.ti.com/lit/wp/spry352/spry352.pdf
[9] Z. Ravichandran et al., "RoboGuard: Safety Guardrails for LLM-Enabled Robots," arXiv preprint arXiv:2503.07885, 2025. [Online]. Available: https://arxiv.org/pdf/2503.07885v1.pdf
[10] Dong, Y., Mu, R., Zhang, Y., Sun, S., Zhang, T., Wu, C., Jin, G., Qi, Y., Hu, J., Meng, J., Bensalem, S., & Huang, X. (2025). Safeguarding large language models: A survey. Artificial Intelligence Review, 58(12), 382. https://doi.org/10.1007/s10462-025-11389-2
[11] Microsoft, "AI-Powered Success: Japan Airlines and Phi-4," Microsoft Cloud Blog, July 2025. [Online]. Available: https://www.microsoft.com/en-us/microsoft-cloud/blog/2025/07/24/ai-powered-success-with-1000-stories-of-customer-transformation-and-innovation/.
