Detection of the frequency-hopping (FH) signal is challenging when the hopping rate is unknown. Conventional spectrogram-based schemes can detect FH signals, but its performance is limited by the time-frequency resolution trade-off and spectral leakage. To alleviate this issue, we design convolutional neural network (CNN) and hybrid CNN/recurrent neural network (RNN)-based schemes. The CNN-based scheme alleviates spectral leakage by using feature maps. The hybrid CNN/RNN-based scheme mitigates the time-frequency resolution trade-off by using feature maps extracted from spectrograms of various window lengths. In simulations, the hybrid CNN/RNN-based scheme is shown to outperform the CNN-based and conventional detection schemes.

The frequency-hopping technique is widely used in commercial and military communications for its superiority in anti-jamming and low probability of interception and detection capabilities. To further lower its detection probability, a frequency-hopping signal randomly changes its carrier frequency within a transmission symbol. Conventional detection methods have poor performance in the presence of fast frequency-hopping signals. To solve this problem, we propose a Dirty-Template-based detection scheme. Dirty Template was originally proposed to blindly estimate the time delay in an ultra-wide-band system. We propose that it can be employed in the frequency domain to detect fast frequency-hopping signal. The decision statistic derived from the cross-correlation between the template and received signals in the frequency domain is analyzed. Using this cross-correlation, a detection scheme, based on the Neyman-Pearson test is proposed. The simulation results show that the Dirty-Template-based scheme outperforms the autocorrelation-based scheme when the hopping period is short.

We study the tradeoff between energy and bandwidth efficiency for short packet-transmission systems for wireless sensor networks. Using random coding bound (RCB) analysis, we show that 16-ary and 32-ary orthogonal modulation schemes provide reasonable balance between energy and bandwidth efficiency. In addition, we compare various channel-coding schemes and conclude that convolutional codes are well suited for wireless sensor network applications having excellent performance and reasonable robustness to imperfections, such as estimation errors.

In this study, we investigate the operation of a user-number threshold-based base station (BS) on/off control, in which the BS turns off when the number of users is less than a specific threshold value. This paper presents a space-based analysis of the BS on/off control system to which a stochastic geometric approach is applied. In particular, we derive the approximated closed-form expression of the coverage probability of a homogeneous network (HomNet) with the user-number threshold-based BS on/off control. Moreover, the optimal user-number threshold for maximizing coverage probability is analytically derived. In addition to HomNet, we also derive the approximated closed-form expression of the coverage probability and the optimal user-number thresholds for a heterogeneous network (HetNet). The results show that HetNet, the analysis of which seems intractable, can be analyzed in the form of a linear combination of HomNets with weighted densities. In addition, the optimal user-number threshold of each tier is obtained independently from other tiers. The modeling and analysis presented in this paper are not only limited to the case of user-number threshold-based on/off control, but also applicable to other novel on/off controls with minor modifications. Finally, by comparing the analytical results with the simulated results, the theoretical contributions of this study are validated.