This letter proposes the error-correctable shaping method based on the trellis diagram defined in enumerative sphere shaping (ESS), refer to trellis-based error-correctable ESS (TE-ESS), to mitigate one of the implementation issues in probabilistic amplitude shaping (PAS). To introduce the residual error detection and correction, TE-ESS modifies the trellis diagram of ESS to restrict the output amplitude sequence which correspond to even- or odd-parity. Numerical simulations show that TE-ESS outperforms the previously proposed error-correctable shaping method. In addition, Its error correction effectively alleviates the error floor requirement in FEC code, compared to using ESS, in low BER region.

To increase data rate of screen camera communication (SCC), rolling shutter camera-based SCCs (RS-SCCs) have been studied. Among modulation schemes, inverse pulse position modulation (I-PPM) is the simplest to maintain constant screen luminance. However, image overlay properties of I-PPM have yet to be elucidated. In this study, the properties were clarified for uplink SCC that combined with a small liquid crystal display and a telephoto RS camera. Type of overlay standard image for I-PPMs and variable PPMs (VPPMs) was altered as parameters. Comparison of symbol error rates among I-PPMs and VPPMs demonstrated the most suitable scheme for image overlay RS-SCC.

For drones to expand their activities, a self-localization method for indoor flying drones is required to complement GPS. We have investigated indoor drone positioning based on Wi-Fi RTT (Round Trip Time). This paper presents methods for estimating the position coordinate of a drone using Wi-Fi RTT and machine learning. In addition to a method that learns actual Wi-Fi RTT ranging data, we propose a novel method that learns pseudo-generated ranging data reproducing Wi-Fi RTT characteristics. Experimental results show that the proposed machine learning-based method using pseudo-generated data achieves higher accuracy than the method that learns actual ranging data and is also superior to the conventional MMSE method.

In the development of large-scale communication software, various methods have been investigated to address challenges such as increasing development costs and shortages of skilled personnel by leveraging machine learning to automatically generate system test cases from requirement specifications. To further improve the accuracy of automatically generated test cases, this study proposes a method that employs deep learning to structure requirement specifications and generate test cases. In particular, we introduce an Attention Seq2Seq model. The proposed model achieved significantly higher accuracy than the LSTM model presented in previous studies.

Grant-free random access, as a crucial access mechanism in current massive machine-type communication (mMTC) scenarios, can significantly reduce signaling overhead. Base station(BS) achieve joint device activity detection and channel estimation (JADCE) through received pilot sequences. However, time delays severely degrade detection performance and increase computational complexity. In this letter, we first design a novel pilot sequence composed of chirp signals with distinct parity index values. Then, we construct a reduced candidate pilot set by leveraging the invariant parity property of index value sums under time delays. Finally, exploiting the favorable correlation characteristics of the pilot sequences, we propose a low-complexity orthogonal matching pursuit with successive interference cancellation (OMP-SIC) algorithm for solving the JADCE problem in asynchronous multiple access scenarios.

Visible light communication (VLC) has emerged as an alternative to data communication, offering higher bandwidth than radio communication. To fully leverage its advantages, Polar codes, recognized as capacity-achieving codes, are employed in serial concatenation with run-length limited (RLL). In this paper, we propose an enhancement to our previous decoding method, successive cancellation list - soft guessing random additive noise decoding (SCL-SGRAND), termed thresholded SCLSGRAND (T-SCLSGRAND). By introducing a variable called the threshold t, which reduces the length of the soft information used as the input for SGRAND in T-SCLSGRAND, this method demonstrates improved bit-error rate (BER) performance compared to successive cancellation list (SCL) and cyclic redundancy check (CRC)-aided SCL (CA-SCL) by 0.7 and 0.2 dB, respectively, under a small upper limit b.