近年來，物聯網架構的探索已成為一項備受關注的研究課題。本研究延伸了[1, 2]中的網內學習框架，該框架利用空中計算的概念，協調物聯網設備以形成一個大規模且可部署於實際場域的等效神經網路。我們探討了截斷通道反轉技術在固定層指派環境中的應用，並聚焦於該框架下傳輸模組的設計。此外，基於網內學習的端到端訓練方法，我們進一步解決了原始層指派訓練設計中的低效問題。為了應對這一挑戰，我們提出了一種群組路由策略，該策略結合了貪婪轉發與分群技術進行預分群。此方法採用逐跳的方式，使得每個當前跳點群組能夠簡單直觀地做出局部跳躍決策。在鄰近區域內，會識別出符合地理進展標準的候選節點，並基於這些候選節點，應用基於密度的分群演算法來構建下一跳的候選群組。作為一種離線預分群方法，這些策略旨在降低初始訓練階段前的能量消耗，並提升訓練效率。模擬結果驗證了我們所提出的方法、貪婪方法及最短路徑方法的可行性，並提供了性能比較，突顯出我們的方法在優化物聯網應用的網內學習架構方面的潛力。

The exploration of IoT architectures has become a prominent research topic in recent years. This work extends the in-network learning framework [1, 2], which coordinates IoT devices to form a large-scale, field-deployed equivalent neural network by leveraging over-the-air communication concepts. We examine the application of the truncated channel inversion scheme in a fixed layer assignment environment, focusing on the design of the transmission module under this framework. Additionally, building on the end-to-end training approach in in-network learning, we address the inefficiencies in the original layer-assignment training design. To tackle this issue, we propose group routing strategies that integrates greedy forwarding with clustering techniques for pre-grouping. In a hop-by-hop manner, the proposed group routing method is simple and intuitive to make local hop decisions at each current hop group. Within the neighborhood, candidates that meet the geographic progress criteria are identified. Starting from these candidates, a density-based clustering algorithm is applied to construct candidate groups for the next hop. These strategies, as an offline pre-grouping method, aim to reduce energy consumption and enhance training efficiency before the initial training phase. Simulation results validate the feasibility of our method, greedy method, and shortest-path method, providing a performance comparison, highlighting our method's potential to optimize in-network learning architectures for IoT applications.

[1] C.-C. Wang, “In-network learning via over-the-air computation in internet-of-things,” Master’s thesis, National Tsing Hua University, 2022.

[2] Y.-W. Peter Hong and C.-C. Wang, “In-network learning via over-the-air computation in Internet-of-Things,” in IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), 2021, pp. 141–145.

[3] K. Evangelos, S. Harvinder, and U. Jorge, “Compass routing on geometric networks,” in Proceedings of the Canadian Conference on Computational Geometry, 1999, pp. 51–54.

[4] P. Bose, P. Morin, I. Stojmenović, and J. Urrutia, “Routing with guaranteed delivery in ad hoc wireless networks,” in International Workshop on Discrete Algorithms and Methods for Mobile Computing and Communications, 1999, pp. 48–55.

[5] B. Karp and H.-T. Kung, “GPSR: Greedy perimeter stateless routing for wireless networks,” in Annual International Conference on Mobile Computing and Networking, 2000, pp. 243–254.

[6] J. Na and C.-K. Kim, “GLR: A novel geographic routing scheme for large wireless ad hoc networks,” Computer Networks, vol. 50, no. 17, pp. 3434–3448, 2006.

[7] D. Zhang and E. Dong, “An efficient bypassing void routing protocol based on virtual coordinate for WSNs,” IEEE Communications Letters, vol. 19, no. 4, pp. 653–656, 2015.

[8] H. Huang, H. Yin, G. Min, J. Zhang, Y. Wu, and X. Zhang, “Energy-aware dual-path geographic routing to bypass routing holes in wireless sensor networks,” IEEE Transactions on Mobile Computing, vol. 17, no. 6, pp. 1339–1352, 2017.

[9] H. Zhang and H. Shen, “Energy-efficient beaconless geographic routing in wireless sensor networks,” IEEE Transactions on Parallel and Distributed Systems, vol. 21, no. 6, pp. 881–896, 2010.

[10] N. T. Hadi and Wibisono, “A swing routing approach to improve performance of shortest geographical routing protocol for wireless sensor networks,” in International Conference on Information and Communications Technology (ICOIACT), 2019, pp. 291–296.

[11] M. Chen, V. C. Leung, S. Mao, and Y. Yuan, “Directional geographical routing for real-time video communications in wireless sensor networks,” Computer Communications, vol. 30, no. 17, pp. 3368–3383, 2007.

[12] J. Wang, Y. Zhang, J. Wang, Y. Ma, and M. Chen, “PWDGR: Pair-wise directional geographical routing based on wireless sensor network,” IEEE Internet of Things Journal, vol. 2, no. 1, pp. 14–22, 2015.

[13] J. MacQueen, “Some methods for classification and analysis of multivariate observations,” in Proceedings of Berkeley Symposium on Mathematical Statistics and Probability, 1967.

[14] D. Reynolds, “Gaussian mixture models,” Encyclopedia of Biometrics, vol. 741, no. 659–663, 2009.

[15] J. C. Bezdek, R. Ehrlich, and W. Full, “FCM: The fuzzy c-means clustering algorithm,” Computers and Geosciences, vol. 10, no. 2–3, pp. 191–203, 1984.

[16] M. Ester, H.-P. Kriegel, J. Sander, and X. Xu, “A density-based algorithm for discovering clusters in large spatial databases with noise,” in Proceedings of the International Conference on Knowledge Discovery and Data Mining, vol. 96, no. 34, 1996, pp. 226–231.

[17] P. Liu, D. Zhou, and N. Wu, “VDBSCAN: varied density based spatial clustering of applications with noise,” in International Conference on Service Systems and Service Management, IEEE, 2007, pp. 1–4.

[18] Z. Cai, J. Wang, and K. He, “Adaptive density-based spatial clustering for massive data analysis,” IEEE Access, vol. 8, pp. 23 346–23 358, 2020.

[19] Z. Wang, Z. Ye, Y. Du, Y. Mao, Y. Liu, Z. Wu, and J. Wang, “AMD-DBSCAN: An adaptive multi-density DBSCAN for datasets of extremely variable density,” in International Conference on Data Science and Advanced Analytics (DSAA), 2022.

[20] K. Sawant, “Adaptive methods for determining DBSCAN parameters,” International Journal of Innovative Science, Engineering & Technology, vol. 1, no. 4, pp. 329–334, 2014.

[21] L. Zhu, C. Li, B. Xia, Y. He, and Q. Lin, “A hybrid routing protocol for 3-D vehicular ad hoc networks,” IEEE Systems Journal, vol. 11, no. 3, pp. 1239–1248, 2017.

[22] R. Manjula, T. Koduru, and R. Datta, “Protecting source location privacy in IoT-enabled wireless sensor networks: The case of multiple assets,” IEEE Internet of Things Journal, vol. 9, no. 13, pp. 10 807–10 820, 2022.

[23] J. Yang, K. Sun, H. He, X. Jiang, and S. Chen, “Dynamic virtual topology aided networking and routing for aeronautical ad-hoc networks,” IEEE Transactions on Communications, vol. 70, no. 7, pp. 4702–4716, 2022.

[24] P. Ponnavaikko, S. K. Wilson, M. Stojanovic, J. Holliday, and K. Yassin, “Delay-constrained energy optimization in high-latency sensor networks,” IEEE Sensors Journal, vol. 17, no. 13, pp. 4287–4298, 2017.

[25] O. Alzamzami and I. Mahgoub, “Fuzzy logic-based geographic routing for urban vehicular networks using link quality and achievable throughput estimations,” IEEE Transactions on Intelligent Transportation Systems, vol. 20, no. 6, pp. 2289–2300, 2019.

[26] B. Fan and Y. Xin, “A clustering and routing algorithm for fast changes of large-scale WSN in IoT,” IEEE Internet of Things Journal, vol. 11, no. 3, pp. 5036–5049, 2024.

[27] M. Goldenbaum and S. Stanczak, “On the channel estimation effort for analog computation over wireless multiple-access channels,” IEEE Wireless Communications Letters, vol. 3, no. 3, pp. 261–264, 2014.

[28] M. Soltanalian, M. M. Naghsh, N. Shariati, P. Stoica, and B. Hassibi, “Training signal design for correlated massive MIMO channel estimation,” IEEE Transactions on Wireless Communications, vol. 16, no. 2, pp. 1135–1143, 2017.