A space–air–ground–sea integrated network (SAGSIN) is a promising heterogeneous network framework for the next generation mobile communications. Moreover, federated learning (FL), as a widely used distributed intelligence approach, can improve advanced network performance. In view of the combination and cooperation of SAGSINs and FL, an FL-based SAGSIN framework faces a number of unprecedented challenges, not only from the communication aspect but also on the security and privacy side. Motivated by these observations, in this article, we first give a detailed state-of-the-art review of recent progress and ongoing research works on FL-based SAGSINs. Then, the challenges of FL-based SAGSINs are discussed. After that, for different service demands, basic applications are introduced with their benefits and functions. In addition, two case studies are proposed, in order to improve SAGSINs’ communication efficiency under a significant communication latency difference and to protect user-level privacy for SAGSIN participants, respectively. Simulation results show the effectiveness of the proposed algorithms. Moreover, future trends of FL-based SAGSINs are discussed.

The propagation delay is non-negligible in underwater acoustic networks (UANs) since the propagation speed is five orders of magnitude smaller than the speed of light. In this case, space and time factors are strongly coupled to determine the collisions of packet transmissions. To this end, this paper analyzes the impact of space-time coupling on slotted medium access control (MAC). We find that a sending node has specific location-dependent interference slots and slot-dependent interference regions. Thus, the collisions may span multiple slots, leading to both inter-slot and intra-slot collisions. Interestingly, the slot-dependent interference regions could be annulus inside the whole transmission range. It is pointed out that collision-free regions exist when a guard interval in a slot is larger than a packet duration. In this sense, the long slot brings spatial reuse within the transmission range. We then derive the closed-form expressions for the successful transmission probability of Slotted-ALOHA and the upper-bound and the lower-bound for the successful transmission probability in UANs. We further find that the optimal guard interval to reach the peak successful transmission probabilities is not larger than a packet duration, which is much shorter than the existing slot setting in the typical Slotted-ALOHA in the uniformly distributed UANs. Simulation results verify our findings, and also show that the performance of vertical transmissions is more sensitive to the spatial impact than horizontal transmissions in UANs.

In the ocean environment, passive acoustic monitoring (PAM) is an important technique for the surveillance of cetacean species. Manual detection for a large amount of PAM data is inefficient and time-consuming. To extract useful features from a large amount of PAM data for classifying different cetacean species, we propose an automatic detection and unsupervised clustering-based classification method for cetacean vocal signals. This paper overcomes the limitations of the traditional threshold-based method, and the threshold is set adaptively according to the mean value of the signal energy in each frame. Furthermore, we also address the problem of the high cost of data training and labeling in deep-learning-based methods by using the unsupervised clustering-based classification method. Firstly, the automatic detection method extracts vocal signals from PAM data and, at the same time, removes clutter information. Then, the vocal signals are analyzed for classification using a clustering algorithm. This method grabs the acoustic characteristics of vocal signals and distinguishes them from environmental noise. We process 194 audio files in a total of 25.3 h of vocal signal from two marine mammal public databases. Five kinds of vocal signals from different cetaceans are extracted and assembled to form 8 datasets for classification. The verification experiments were conducted on four clustering algorithms based on two performance metrics. The experimental results confirm the effectiveness of the proposed method. The proposed method automatically removes about 75% of clutter data from 1581.3MB of data in audio files and extracts 75.75 MB of the features detected by our algorithm. Four classical unsupervised clustering algorithms are performed on the datasets we made for verification and obtain an average accuracy rate of 84.83%.

Many marine applications are required to obtain data from underwater sources and transmit it to a data center situated on the water or land. The data transmissions must cross the water-air interface using a surface gateway. In this process, the gateway traditionally demodulates underwater acoustic (UWA) signals and forwards them via a radio link. This paper proposes a novel cloud software-defined acoustic (C-SDA) framework for cross-interface relaying. In contrast to the conventional approach of demodulating UWA signals at the gateway, the C-SDA framework employs a sampling and forward (SF) relaying strategy. This method only samples the passband UWA signals, effectively circumventing the necessity for dual-stack packet encapsulation and re-encapsulation at the gateway. Since passband sampling may result in a large amount of data to the radio link, SF relaying uses radio communication bandwidth to exchange cloud computation resources. C-SDA also implements softwarization of the UWA transceiver functions in the cloud (i.e., the monitoring center). Thus, the challenging UWA channel can be addressed by the advanced but complex algorithms in the cloud. C-SDA not only simplifies the design of the gateway but also facilitates rapid technique updates and iterations for the UWA transceiver. The field-test experimental results show that the proposed C-SDA can enable complex equalizers and achieve significant bit error rate improvement compared to the self-developed embedded UWA receiver. Additionally, the future research directions for C-SDA are discussed in the paper.

Learning-based equalizers for multicarrier communication systems have been widely studied over underwater acoustic (UWA) channels. In this article, a learning-based equalizer is utilized for single-carrier (SC) underwater acoustic communications. A comprehensive comparison is made between existing deep learning (DL)-based approaches and a classical equalizer designed with adaptive filtering principles. It motivates the design of equalization for SC communications over underwater acoustic channels. To overcome distortion over the UWA channel, we propose a sliding deep learning-based equalizer that uses a sliding nonlinear network for equalization rather than a single-layer linear method. Moreover, to accelerate convergence during training, we proposed a preprocessing-based training phase. To mitigate the impact of time-varying channels, we additionally propose a meta-learning-enhanced adaptive filter algorithm for online adaptive equalization, named Meta-DNN. Based on the proposed DL equalizer, we leverage the pilot and data relationship to perform online transfer to achieve better BER performance. Moreover, to make this work more convincing, we test bit-error-rate (BER) performance across reproducible, realistic multi-scenario channels.

Due to the space–time coupling access, we find that anti-eavesdropping opportunities exist in underwater acoustic networks (UANs), where packets can be successfully received only by the intended receiver, but collide at the unintended receivers. These opportunities are highly spatially dependent, and this paper studies the case that linearly deployed sensor nodes directly report data toward a single collector. We found an eavesdropping ring centered around these linearly deployed sensor nodes, where the eavesdropper could steal all the reported data. Since the typical receiving-alignment-based scheduling MAC (TRAS-MAC) will expose the relative spatial information among the sensor nodes with the collector, the eavesdropper can locate the eavesdropping ring. Although moving the collector into the one-dimensional sensor node chain can degrade the eavesdropping ring to a point that constrains the eavesdropping risk, the collector’s location will be subsequently exposed to the eavesdropper. To efficiently protect the reported data and prevent the exposure of the collector’s location, we designed a slotted- and receiving-alignment-based scheduling MAC (SRAS-MAC). The NS-3 simulation results showed the effectiveness of the SRAS-MAC and the TRAS-MAC in protecting data from eavesdropping, which protect 90% of the data from eavesdropping in the one-eavesdropper case and up to 80% of data from eavesdropping in ten-eavesdropper cases. Moreover, unlike TRAS-MAC, which will expose the collector’s location, SRAS-MAC provides multiple positions for the collector to hide, and the eavesdropper cannot distinguish where it is.