Revisiting Non-Coherent Detection of Differential PSK: A Factor Graph Perspective

Shi, Long; Wang, Taotao; Mei, Zhen; Li, Jun; Xia, Xiang-Gen

doi:10.1109/lcomm.2022.3231920

Cited by 2 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As mentioned in Refs. 50 and 51, the patch tokens tend to be smoothed as the transformer blocks are stacked, which can have a significant impact on the final segmentation performance. To address this issue, we add an auxiliary classification layer to the intermediate for classification and generating auxiliary CAMs.…”

Section: Methodsmentioning

confidence: 99%

TSD-CAM: transformer-based self distillation with CAM similarity for weakly supervised semantic segmentation

Yan,

Chen,

Tang

2024

J. Electron. Imag.

View full text Add to dashboard Cite

Weakly supervised semantic segmentation (WSSS) using only image-level labels is a challenging task. Most existing methods utilize class activation map (CAM) to generate pixel-level pseudo labels for supervised training. However, the gap between classification and segmentation hinders the network from obtaining more comprehensive semantic information and generating more accurate pseudo masks for segmentation. To address this issue, we propose TSD-CAM, a transformer-based self distillation (SD) method that utilizes CAM similarity. TSD-CAM uses the similarity between CAMs generated from different views as a distillation target, providing additional supervision for the network and narrowing the gap between classification and segmentation. SD supervision allows the network to acquire more semantic information and refine CAMs to generate higher precision pseudo-labels. In addition, we propose the adaptive pixel refinement module, which adaptively refines and adjusts images based on pixel variations, further improving the precision of pseudo labels. Our method is a fully end-to-end single-stage approach that achieves stateof-the-art 71.3% mIoU on PASCAL VOC 2012 and 42.9% mIoU on the MS COCO 2014 dataset, and the proposed TSD-CAM can significantly outperform other singlestage competitors and achieve comparable performance with state-of-the-art multistage methods. Meanwhile, the effectiveness of our method is demonstrated by a large number of ablation experiments, and we provide a new way of thinking to solve the problems of WSSS. Our code is available at: https://github.com/pipizhum/ TSD-CAM.

show abstract

Section: Methodsmentioning

confidence: 99%

TSD-CAM: transformer-based self distillation with CAM similarity for weakly supervised semantic segmentation

Yan,

Chen,

Tang

2024

J. Electron. Imag.

View full text Add to dashboard Cite

show abstract

“…The ranges are given by 0 ≤ l p < ⌈M ∆f τ max ⌉ and −⌈N T ϑ max ⌉ < k p < ⌈N T ϑ max ⌉. Therefore, each propagation path seen in ( 2) can be uniquely identified by the DD-domain parameters of fading gain h p , Doppler index k p = kp + δk p and delay index l p = lp + δl p , 1 It has been recently reported in [42] that the TF-domain MP detection can be applied to differential schemes in a random-walk channel phase model, where the factorization of the joint probability function of the noise and channel phase is exploited. However, this cannot be applied to the OTFS regime, where the time-varying dynamics of the TF-domain fading channel are determined by the Doppler frequency instead of a random channel phase.…”

Section: Coherent Otfs Relying On Practical Csi Estimation a General ...mentioning

confidence: 99%

“…Given the dimensions of y β ∈ C M N ×1 in (35), there is only one non-zero eigenvalue for the noncoherent OTFS using the V-BLAST model. In summary, upon solving (42) based on calculating the correlation metrices of ( 40) and ( 41), the filtering weights w κ is obtained for (36) without CSI.…”

Section: A Noncoherent Otfs System Design Based On Differential V-blastmentioning

confidence: 99%

Noncoherent Orthogonal Time Frequency Space Modulation

Xu,

Xiang,

Sugiura

et al. 2024

IEEE Trans. Wireless Commun.

View full text Add to dashboard Cite

The recently-developed orthogonal time frequency space (OTFS) modulation is capable of transforming the timevarying fading of the time-frequency (TF) domain into the time-invariant fading representations of the delay-Doppler (DD) domain. The OTFS system using orthogonal frequency-division multiplexing (OFDM) as inner core naturally requires the subcarrier spacing (SCS) ∆f to be larger than the maximum Doppler frequency ϑmax, i.e. ∆f > ϑmax, when perfect channel state information (CSI) knowledge is assumed. However, for the first time in literature, we explicitly demonstrate that the practical OFDM-based OTFS systems have to double their SCS in order to facilitate CSI estimation, requiring ∆f ′ = 2∆f > 2ϑmax.In order to mitigate this loss, we propose a novel noncoherent OTFS system, which is capable of operating at ∆f > ϑmax. The major challenge in this context is the mitigation of the DDdomain interference without CSI. Against this background, we draw an analogy between the input-output model of OTFS and that of V-BLAST, where V-BLAST's blind inter-antenna interference mitigation technique is invoked. Moreover, we propose to partition the DD-domain modulated symbols into groups, where space-time block coding is invoked in order to eliminate the DDdomain interference within each group. Our simulation results demonstrate that the proposed noncoherent OTFS is capable of substantially outperforming its coherent counterparts relying on CSI estimation.

show abstract

Revisiting Non-Coherent Detection of Differential PSK: A Factor Graph Perspective

Cited by 2 publications

References 15 publications

TSD-CAM: transformer-based self distillation with CAM similarity for weakly supervised semantic segmentation

TSD-CAM: transformer-based self distillation with CAM similarity for weakly supervised semantic segmentation

Noncoherent Orthogonal Time Frequency Space Modulation

Contact Info

Product

Resources

About