--- title: MobiSys 2024 Digest venue: MobiSys year: 2024 date: '2024-01-01' tags: [] paper_count: 13 draft: false --- 13 papers selected. --- ### WAIS: Leveraging WiFi for Resource-Efficient SLAM *Aditya Arun 0002, William Hunter, Roshan Sai Ayyalasomayajula, Dinesh Bharadia* **TL;DR** — Demonstrates that commodity WiFi signals can replace LiDAR for simultaneous localization and mapping, dramatically cutting the resource cost of robot/AR navigation. --- ### UWB-Fi: Pushing Wi-Fi towards Ultra-wideband for Fine-Granularity Sensing *Xin Li 0070, Hongbo Wang, Zhe Chen 0015, Zhiping Jiang *et al.** **TL;DR** — Extends standard Wi-Fi to UWB-class sensing resolution without hardware changes, enabling centimeter-level gesture and motion detection on existing infrastructure. --- ### Radarize: Enhancing Radar SLAM with Generalizable Doppler-Based Odometry *Emerson Sie, Xinyu Wu, Heyu Guo, Deepak Vasisht* **TL;DR** — Introduces a Doppler-derived odometry method that generalizes radar-based SLAM across environments and radar hardware without per-deployment retraining. --- ### ChirpTransformer: Versatile LoRa Encoding for Low-power Wide-area IoT *Chenning Li, Yidong Ren, Shuai Tong, Shakhrul Iman Siam *et al.** **TL;DR** — Redesigns LoRa chirp encoding with a transformer-based scheme that simultaneously improves throughput, range, and coexistence for large-scale IoT deployments. --- ### Willow: Practical WiFi Backscatter Localization with Parallel Tags *Jinyan Jiang, Jiliang Wang, Yijie Chen, Shuai Tong *et al.** **TL;DR** — Enables concurrent localization of multiple passive backscatter tags over commodity WiFi, making large-scale battery-free asset tracking practical. --- ### Pantheon: Preemptible Multi-DNN Inference on Mobile Edge GPUs *Lixiang Han, Zimu Zhou, Zhenjiang Li* **TL;DR** — Provides a preemptible scheduling runtime for concurrent DNN workloads on edge GPUs, achieving low-latency inference without sacrificing throughput under mixed real-time demands. --- ### ARISE: High-Capacity AR Offloading Inference Serving via Proactive Scheduling *Z. Jonny Kong, Qiang Xu 0006, Y. Charlie Hu* **TL;DR** — Proactively schedules AR inference offloading by predicting gaze and scene dynamics, significantly increasing server capacity while meeting strict latency budgets. --- ### CACTUS: Dynamically Switchable Context-aware micro-Classifiers for Efficient IoT Inference *Mohammad Mehdi Rastikerdar, Jin Huang, Shiwei Fang, Hui Guan 0001 *et al.** **TL;DR** — Deploys a family of tiny context-aware classifiers on microcontrollers that switch at runtime to match workload context, cutting energy by orders of magnitude versus monolithic models. --- ### Empowering In-Browser Deep Learning Inference on Edge Through Just-In-Time Kernel Optimization *Fucheng Jia, Shiqi Jiang 0002, Ting Cao 0003, Wei Cui *et al.** **TL;DR** — Uses JIT kernel specialization to close the performance gap between browser-based and native DNN inference on edge devices, enabling high-throughput on-device AI in web apps. --- ### FedConv: A Learning-on-Model Paradigm for Heterogeneous Federated Clients *Leming Shen, Qiang Yang 0018, Kaiyan Cui, Yuanqing Zheng *et al.** **TL;DR** — Proposes learning directly over model parameters rather than data, allowing federated learning to work across radically heterogeneous IoT devices without sharing raw data or requiring uniform architectures. --- ### SoilCares: Towards Low-cost Soil Macronutrients and Moisture Monitoring Using RF-VNIR Sensing *Juexing Wang, Yuda Feng, Gouree Kumbhar, Guangjing Wang 0001 *et al.** **TL;DR** — Combines RF and near-infrared sensing in a low-cost handheld device to measure soil nutrients and moisture, demonstrating real agricultural field deployments. --- ### MobiAir: Unleashing Sensor Mobility for City-scale and Fine-grained Air-Quality Monitoring with AirBERT *Yuxuan Liu 0010, Haoyang Wang 0012, Fanhang Man, Jingao Xu *et al.** **TL;DR** — Leverages mobile sensors on vehicles and pedestrians with a BERT-style spatio-temporal model to achieve city-scale, fine-grained air quality maps at a fraction of the cost of static sensor networks. --- ### Joey: Supporting Kangaroo Mother Care with Computational Fabrics *Qijia Shao, Jiting Liu, Emily Bejerano, Ho-Man Colman Leung *et al.** **TL;DR** — Embeds soft physiological sensors directly into a wearable fabric wrap to monitor premature infants during skin-to-skin care, demonstrating a compelling real-world clinical deployment.