Survey on Near-Space Information Networks: Channel Modeling, Networking, and Transmission Perspectives

Near-space information networks (NSIN) composed of high-altitude platforms (HAPs), high- and low-altitude unmanned aerial vehicles (UAVs) are a new regime for providing quickly, robustly, and cost-efficiently sensing and communication services. Precipitated by innovations and breakthroughs in manufacturing, materials, communications, electronics, and control technologies, NSIN have emerged as an essential component of the emerging sixth-generation of mobile communication systems. This article aims at providing and discussing the latest advances in NSIN in the research areas of channel modeling, networking, and transmission from a forward-looking, comparative, and technological evolutionary perspective. In this article, we highlight the characteristics of NSIN and present the promising use-cases of NSIN. The impact of airborne platforms' unstable movements on the phase delays of onboard antenna arrays with diverse structures is mathematically analyzed. The recent advancements in HAP channel modeling are elaborated on, along with the significant differences between HAP and UAV channel modeling. A comprehensive review of the networking technologies of NSIN in network deployment, handoff management, and network management aspects is provided. Besides, the promising technologies and communication protocols of the physical layer, medium access control (MAC) layer, network layer, and transport layer of NSIN for achieving efficient transmission over NSIN are overviewed. Finally, we outline some open issues and promising directions of NSIN deserved for future study and discuss the corresponding challenges.

Heterogeneous Generative Knowledge Distillation with Masked Image Modeling

Small CNN-based models usually require transferring knowledge from a large model before they are deployed in computationally resource-limited edge devices. Masked image modeling (MIM) methods achieve great success in various visual tasks but remain largely unexplored in knowledge distillation for heterogeneous deep models. The reason is mainly due to the significant discrepancy between the Transformer-based large model and the CNN-based small network. In this paper, we develop the first Heterogeneous Generative Knowledge Distillation (H-GKD) based on MIM, which can efficiently transfer knowledge from large Transformer models to small CNN-based models in a generative self-supervised fashion. Our method builds a bridge between Transformer-based models and CNNs by training a UNet-style student with sparse convolution, which can effectively mimic the visual representation inferred by a teacher over masked modeling. Our method is a simple yet effective learning paradigm to learn the visual representation and distribution of data from heterogeneous teacher models, which can be pre-trained using advanced generative methods. Extensive experiments show that it adapts well to various models and sizes, consistently achieving state-of-the-art performance in image classification, object detection, and semantic segmentation tasks. For example, in the Imagenet 1K dataset, H-GKD improves the accuracy of Resnet50 (sparse) from 76.98% to 80.01%.

Representation Disparity-aware Distillation for 3D Object Detection

In this paper, we focus on developing knowledge distillation (KD) for compact 3D detectors. We observe that off-the-shelf KD methods manifest their efficacy only when the teacher model and student counterpart share similar intermediate feature representations. This might explain why they are less effective in building extreme-compact 3D detectors where significant representation disparity arises due primarily to the intrinsic sparsity and irregularity in 3D point clouds. This paper presents a novel representation disparity-aware distillation (RDD) method to address the representation disparity issue and reduce performance gap between compact students and over-parameterized teachers. This is accomplished by building our RDD from an innovative perspective of information bottleneck (IB), which can effectively minimize the disparity of proposal region pairs from student and teacher in features and logits. Extensive experiments are performed to demonstrate the superiority of our RDD over existing KD methods. For example, our RDD increases mAP of CP-Voxel-S to 57.1% on nuScenes dataset, which even surpasses teacher performance while taking up only 42% FLOPs.

QoE-Driven Video Transmission: Energy-Efficient Multi-UAV Network Optimization

This paper is concerned with the issue of improving video subscribers' quality of experience (QoE) by deploying a multi-unmanned aerial vehicle (UAV) network. Different from existing works, we characterize subscribers' QoE by video bitrates, latency, and frame freezing and propose to improve their QoE by energy-efficiently and dynamically optimizing the multi-UAV network in terms of serving UAV selection, UAV trajectory, and UAV transmit power. The dynamic multi-UAV network optimization problem is formulated as a challenging sequential-decision problem with the goal of maximizing subscribers' QoE while minimizing the total network power consumption, subject to some physical resource constraints. We propose a novel network optimization algorithm to solve this challenging problem, in which a Lyapunov technique is first explored to decompose the sequential-decision problem into several repeatedly optimized sub-problems to avoid the curse of dimensionality. To solve the sub-problems, iterative and approximate optimization mechanisms with provable performance guarantees are then developed. Finally, we design extensive simulations to verify the effectiveness of the proposed algorithm. Simulation results show that the proposed algorithm can effectively improve the QoE of subscribers and is 66.75\% more energy-efficient than benchmarks.

Q-YOLO: Efficient Inference for Real-time Object Detection

Real-time object detection plays a vital role in various computer vision applications. However, deploying real-time object detectors on resource-constrained platforms poses challenges due to high computational and memory requirements. This paper describes a low-bit quantization method to build a highly efficient one-stage detector, dubbed as Q-YOLO, which can effectively address the performance degradation problem caused by activation distribution imbalance in traditional quantized YOLO models. Q-YOLO introduces a fully end-to-end Post-Training Quantization (PTQ) pipeline with a well-designed Unilateral Histogram-based (UH) activation quantization scheme, which determines the maximum truncation values through histogram analysis by minimizing the Mean Squared Error (MSE) quantization errors. Extensive experiments on the COCO dataset demonstrate the effectiveness of Q-YOLO, outperforming other PTQ methods while achieving a more favorable balance between accuracy and computational cost. This research contributes to advancing the efficient deployment of object detection models on resource-limited edge devices, enabling real-time detection with reduced computational and memory overhead.

DCP-NAS: Discrepant Child-Parent Neural Architecture Search for 1-bit CNNs

Neural architecture search (NAS) proves to be among the effective approaches for many tasks by generating an application-adaptive neural architecture, which is still challenged by high computational cost and memory consumption. At the same time, 1-bit convolutional neural networks (CNNs) with binary weights and activations show their potential for resource-limited embedded devices. One natural approach is to use 1-bit CNNs to reduce the computation and memory cost of NAS by taking advantage of the strengths of each in a unified framework, while searching the 1-bit CNNs is more challenging due to the more complicated processes involved. In this paper, we introduce Discrepant Child-Parent Neural Architecture Search (DCP-NAS) to efficiently search 1-bit CNNs, based on a new framework of searching the 1-bit model (Child) under the supervision of a real-valued model (Parent). Particularly, we first utilize a Parent model to calculate a tangent direction, based on which the tangent propagation method is introduced to search the optimized 1-bit Child. We further observe a coupling relationship between the weights and architecture parameters existing in such differentiable frameworks. To address the issue, we propose a decoupled optimization method to search an optimized architecture. Extensive experiments demonstrate that our DCP-NAS achieves much better results than prior arts on both CIFAR-10 and ImageNet datasets. In particular, the backbones achieved by our DCP-NAS achieve strong generalization performance on person re-identification and object detection.

Bi-ViT: Pushing the Limit of Vision Transformer Quantization

Vision transformers (ViTs) quantization offers a promising prospect to facilitate deploying large pre-trained networks on resource-limited devices. Fully-binarized ViTs (Bi-ViT) that pushes the quantization of ViTs to its limit remain largely unexplored and a very challenging task yet, due to their unacceptable performance. Through extensive empirical analyses, we identify the severe drop in ViT binarization is caused by attention distortion in self-attention, which technically stems from the gradient vanishing and ranking disorder. To address these issues, we first introduce a learnable scaling factor to reactivate the vanished gradients and illustrate its effectiveness through theoretical and experimental analyses. We then propose a ranking-aware distillation method to rectify the disordered ranking in a teacher-student framework. Bi-ViT achieves significant improvements over popular DeiT and Swin backbones in terms of Top-1 accuracy and FLOPs. For example, with DeiT-Tiny and Swin-Tiny, our method significantly outperforms baselines by 22.1% and 21.4% respectively, while 61.5x and 56.1x theoretical acceleration in terms of FLOPs compared with real-valued counterparts on ImageNet.

Energy-Efficient URLLC Service Provision via a Near-Space Information Network

The integration of a near-space information network (NSIN) with the reconfigurable intelligent surface (RIS) is envisioned to significantly enhance the communication performance of future wireless communication systems by proactively altering wireless channels. This paper investigates the problem of deploying a RIS-integrated NSIN to provide energy-efficient, ultra-reliable and low-latency communications (URLLC) services. We mathematically formulate this problem as a resource optimization problem, aiming to maximize the effective throughput and minimize the system power consumption, subject to URLLC and physical resource constraints. The formulated problem is challenging in terms of accurate channel estimation, RIS phase alignment, theoretical analysis, and effective solution. We propose a joint resource allocation algorithm to handle these challenges. In this algorithm, we develop an accurate channel estimation approach by exploring message passing and optimize phase shifts of RIS reflecting elements to further increase the channel gain. Besides, we derive an analysis-friend expression of decoding error probability and decompose the problem into two-layered optimization problems by analyzing the monotonicity, which makes the formulated problem analytically tractable. Extensive simulations have been conducted to verify the performance of the proposed algorithm. Simulation results show that the proposed algorithm can achieve outstanding channel estimation performance and is more energy-efficient than diverse benchmark algorithms.

Confidence-driven Bounding Box Localization for Small Object Detection

Despite advancements in generic object detection, there remains a performance gap in detecting small objects compared to normal-scale objects. We for the first time observe that existing bounding box regression methods tend to produce distorted gradients for small objects and result in less accurate localization. To address this issue, we present a novel Confidence-driven Bounding Box Localization (C-BBL) method to rectify the gradients. C-BBL quantizes continuous labels into grids and formulates two-hot ground truth labels. In prediction, the bounding box head generates a confidence distribution over the grids. Unlike the bounding box regression paradigms in conventional detectors, we introduce a classification-based localization objective through cross entropy between ground truth and predicted confidence distribution, generating confidence-driven gradients. Additionally, C-BBL describes a uncertainty loss based on distribution entropy in labels and predictions to further reduce the uncertainty in small object localization. The method is evaluated on multiple detectors using three object detection benchmarks and consistently improves baseline detectors, achieving state-of-the-art performance. We also demonstrate the generalizability of C-BBL to different label systems and effectiveness for high resolution detection, which validates its prospect as a general solution.

Q-ViT: Accurate and Fully Quantized Low-bit Vision Transformer

The large pre-trained vision transformers (ViTs) have demonstrated remarkable performance on various visual tasks, but suffer from expensive computational and memory cost problems when deployed on resource-constrained devices. Among the powerful compression approaches, quantization extremely reduces the computation and memory consumption by low-bit parameters and bit-wise operations. However, low-bit ViTs remain largely unexplored and usually suffer from a significant performance drop compared with the real-valued counterparts. In this work, through extensive empirical analysis, we first identify the bottleneck for severe performance drop comes from the information distortion of the low-bit quantized self-attention map. We then develop an information rectification module (IRM) and a distribution guided distillation (DGD) scheme for fully quantized vision transformers (Q-ViT) to effectively eliminate such distortion, leading to a fully quantized ViTs. We evaluate our methods on popular DeiT and Swin backbones. Extensive experimental results show that our method achieves a much better performance than the prior arts. For example, our Q-ViT can theoretically accelerates the ViT-S by 6.14x and achieves about 80.9% Top-1 accuracy, even surpassing the full-precision counterpart by 1.0% on ImageNet dataset. Our codes and models are attached on https://github.com/YanjingLi0202/Q-ViT