TASK-ORIENTED SEMANTIC COMMUNICATION FOR SMART CLASSROOM IMAGE ANALYTICS
Keywords:
Educational big data, Smart classroom, Semantic communication, Image classificationAbstract
Smart classrooms continuously generate large volumes of visual data that enable educational big data analytics, including learner-state assessment and classroom activity understanding. In practice, however, bandwidth-constrained and noise-impaired wireless links render conventional compress-then-transmit pipelines inefficient: pixel-fidelity-oriented compression typically requires high transmission rates to sustain downstream recognition performance. This paper studies a task-oriented semantic communication framework for image analytics over additive white Gaussian noise (AWGN) channels. We develop a unified multi-SNR, multi-rate training and evaluation protocol and benchmark a learned semantic link against a conventional DCT-based link under matched rate budgets. The proposed semantic link integrates (i) an encoder that extracts task-relevant latent representations, (ii) a rate-controllable channel-selection bottleneck that regulates the transmitted feature budget, and (iii) a decoder that reconstructs images for a fixed downstream classifier. Using CIFAR-10 as a reproducible testbed, we report task performance alongside perceptual quality metrics across a grid of SNR and rate settings. Experimental results indicate that the semantic link consistently sustains higher classification accuracy in low-to-medium SNR and low-rate regimes. In addition, PSNR/SSIM do not necessarily exceed the DCT baseline, revealing a task-perception mismatch that favors task-driven transmission. Overall, the proposed framework offers a practical methodology for designing communication pipelines that better support educational big data image analytics.References
[1] Shao J, Mao Y, Zhang J. Learning Task-Oriented Communication for Edge Inference: An Information Bottleneck Approach. IEEE Journal on Selected Areas in Communications, 2022, 40(1): 197–211. DOI: 10.1109/JSAC.2021.3126087.
[2] Shao J, Mao Y, Zhang J. Task-Oriented Communication for Multidevice Cooperative Edge Inference. IEEE Transactions on Wireless Communications, 2023, 22(1): 73–87. DOI: 10.1109/TWC.2022.3191118.
[3] Wang Y, Guo S, Deng Y, Zhang H, Fang Y. Privacy-Preserving Task-Oriented Semantic Communications Against Model Inversion Attacks. IEEE Transactions on Wireless Communications, 2024, 23(8): 10150–10165. DOI: 10.1109/TWC.2024.3369170.
[4] Xie H, Qin Z, Li GY, Juang BH. Deep Learning Enabled Semantic Communication Systems. IEEE Transactions on Signal Processing, 2021, 69: 2663–2675. DOI: 10.1109/TSP.2021.3071210.
[5] Yang W, Du H, Liew ZQ, Lim WYB, Xiong Z, Niyato D, Chi X, Shen X, Miao C. Semantic Communications for Future Internet: Fundamentals, Applications, and Challenges. IEEE Communications Surveys & Tutorials, 2023, 25(1): 213–250. DOI: 10.1109/COMST.2022.3223224.
[6] Liu Y, Wang X, Ning Z, Zhou MC, Guo L, Jedari B. A survey on semantic communications: Technologies, solutions, applications and challenges. Digital Communications and Networks, 2024, 10(3): 528–545. DOI: 10.1016/j.dcan.2023.05.010.
[7] Iyer S, Khanai R, Torse D, Pandya RJ, Rabie KM, Pai K, Khan WU, Fadlullah Z. A Survey on Semantic Communications for Intelligent Wireless Networks. Wireless Personal Communications, 2023, 129(1): 569–611. DOI: 10.1007/s11277-022-10111-7.
[8] Zhang W, et al. Predictive and Adaptive Deep Coding for Wireless Image Transmission in Semantic Communication. IEEE Transactions on Wireless Communications, 2023, 22(8): 5486–5501. DOI: 10.1109/TWC.2023.3234408.
[9] Wu H, Shao Y, Bian C, Mikolajczyk K, Gündüz D. Deep Joint Source-Channel Coding for Adaptive Image Transmission Over MIMO Channels. IEEE Transactions on Wireless Communications, 2024, 23(10): 15002–15017. DOI: 10.1109/TWC.2024.3422794.
[10] Feng J, Cao G, Liu J, Wang Z, Liu Z, Ding Y. Deep joint source-channel coding for wireless image transmission with auxiliary semantics. Physical Communication, 2025, 73: 102836. DOI: 10.1016/j.phycom.2025.102836.