This article investigates several variants of noncoherent multiple-frequency-shift keying (MFSK) that uses Polar coding and evaluates their applicability to robust underwater acoustic underwater communication. A simplified data model is assumed, yielding a low-complexity MFSK symbol estimator, and symbol likelihoods are calculated using sparse pilot symbols. Results presented from field tests in the Stockholm archipelago indicate that a Polar code will decode with a low frame error rate (FER) if not signal-to-noise ratio (SNR) limited and that its performance is superior to convolutional coding. The reliability of short frames, with 0.16–0.5 s transmission time, and long frames, with spectral efficiency up to 0.1 b/s/Hz, are evaluated in the five replay channels of the Watermark benchmark over varying SNR and SNR per bit, . Synchronization is solved with a sliding hypothesis approach and the results show a FER for Watermark channels, where the robustness is significantly improved by either increasing the frequency separation of tones or lowering the code rate and extending the frame length. The performance compares favorably to methods found in the literature, particularly for the less-benign channels of the benchmark. 

A method that defines the superset of Mary permutation frequency-shift keying (SPFSK) alphabets as the symbol space is investigated as a possible robust data link for the underwater acoustic (UWA) channel. The alphabet is equivalent to on-off keying for M=1. Polar coding and time/frequency guards are used for redundancy. The bit error probability of an SPFSK symbol is evaluated to construct the Polar code; it is found to be uniform across bit positions in the label by using natural labeling due to a symmetric fractal behavior in the pairwise symbol error probability, and the heuristic code construction method is directly applicable. A Ricean noncoherent data model is used, and a reduced complexity algorithm for calculating binary log-likelihoods is provided. The method is simulated in an additive white Gaussian noise channel to find a good Polar code construction and evaluate different alphabet parameterizations; the results show that the receiver statistics improve with the alphabet size. SPFSK is further evaluated in the public Watermark benchmark, using both genie- and pilot-assisted parameter tracking, as well as perfect and nonperfect synchronization. Results from the Watermark replay channels NOF1, NCS1, and BCH1 show that ≲ 1% frame error rate is attained at 16≤ b/N0 ≤ 23 dB depending on the channel, using ∼ 0.2 bit/s/Hz spectral efficiency, indicating its potential as a robust data link. Performance in the UWA channel increases substantially by using many tones, and the statistical model must be sufficiently accurate to leverage the processing gain of an SPFSK alphabet with higher dimensionality.

The underwater domain is an environment hostile to humans due to the hydrostatic pressure that rapidly increases with water depth, which has led to underwater robotics becoming an emerging technological field with many commercial-, environmental-, and security-related applications. A major challenge to untethered autonomous underwater vehicles (AUVs) is communicating robot-to-robot and robot-to-topside operator since it must, in most cases, be done acoustically. Meanwhile, the underwater acoustic (UWA) channel is widely considered one of nature's most difficult communication mediums due to the limited frequency range, complicated sound propagation physics, prolonged- and time-varying multipath, and, in some situations, non-Gaussian background noise. The wide variety of UWA channels observed in different locations, and in the same location at different times, also poses a challenge to the research methodology since sea experiments become inherently difficult to repeat. However, replay simulation of recorded channels using the public benchmark Watermark allows direct comparison between publications and is employed extensively in this thesis, complemented by sea experiments for verifying the internal validity of simulation results.

No link solution is perfect for all channels, and a toolbox consisting of methods with varying information rates and robustness is necessary for an adaptive network to exploit the full capability of the channel encountered in situ. A link is called channel-agnostic if its robustness is limited by the ratio of signal power to noise (SNR), rather than the channel characteristics, thereby being robust to many channels one might encounter. The proven information rates for such link methods are relatively low; this, together with recent advancements in other areas, such as error correction codes and joint synchronisation and decoding, motivates the focus on noncoherent methods in this thesis. The second focus is link adaptation and the necessary mechanisms for its implementation in adaptive UWA networks.

The thesis provides a background on various approaches to acoustic signalling, both coherent and noncoherent, and other key components of a noncoherent UWA link, such as symbol alphabets, receiver data models, error correction codes, and time-Doppler synchronisation. The topic of link adaptation in a UWA network is discussed, as is the methodology for research in UWA communication. The included papers provide a set of channel-agnostic link methods with spectral efficiencies in the range 0.02-0.22 (bit/s/Hz), with varying requirements on the SNR and the length of the communication frame, which are enabled by the presented improvements to link methods. Using a Rice-fading model for soft decoding, the robustness to parameter time variation is found to increase substantially by limiting the SNR of the likelihood parametrisation; this result applies to all methods that employ frequency shift keying (FSK). Furthermore, a novel noncoherent symbol alphabet with 1 (bit/s/Hz) maximum spectral efficiency is presented, whose dimensionality M increases the soft decoder performance, specialising to on-off keying (OOK) for M=1. A joint synchronisation and decoding framework is proposed, allowing robust time-Doppler detection with low overhead; its viability is demonstrated in an adverse shallow-water channel with relative platform velocities in the range +/- 4 (m/s). Moreover, a framework for efficient evaluation of link adaptation algorithms is presented, and a link-adaptive ad-hoc UWA network using low-latency implicit feedback is demonstrated through sea experiments.

The research presented herein has been conducted as part of the Swedish Maritime Robotics Centre (SMaRC), a national cross-disciplinary research centre funded by the Swedish Foundation for Strategic Research (SSF). 

Undervattensdomänen är ogästvänlig miljö för människor på grund av det hydrostatiska trycket som snabbt ökar med vattendjupet, vilket har lett till att undervattensrobotik blivit ett växande teknologiområde med tillämpningar inom näringsliv, miljö, och säkerhet. En stor utmaning för kabellösa autonoma undervattensfarkoster är att kommunikationen mellan antingen två drönare, eller mellan drönare och operatör på ytan, i de flesta fallen måste ske akustiskt. Samtidigt är undervattenskanalen allmänt känd som en av naturens svåraste kommunikationsmedium, på grund av begränsat frekvensområdet, kompicerad ljudpropageringsfysik, lång och tidsvarierande flervägsutbredning, samt i vissa situationer, icke-Gaussiskt bakgrundsbrus. Den stora variationen hos undervattenskanaler som observerats på olika platser, och på samma plats men vid olika tidpunkter, är dessutom en utmaning för forskningsmetodologin eftersom sjöförsök i sin natur är svåra att återupprepa. Simuleringar i inspelade kanaler med hjälp av det allmänt tillgängliga verktyget Watermark möjliggör en jämförelse mellan olika publikationer och har därför använts genomgående i denna avhandling, komplementerat med sjöförsök för att validera den interna validiteten hos simuleringsresultaten.

Ingen länklösning passar perfekt till alla kanaler, och en verktygslåda av metoder med olika informationshastighet och robusthet behövs för att ett adaptivt nätverk ska kunna utnyttja den observerade undervattenkanalens fulla potential. En länk kallas kanaloberoende om dess robusthet inte beror på kanalens egenskaper, utan enbart begränsas av förhållandet mellan signal- och bruseffekt. Den påvisade informationshastigheten för sådana länkmetoder är relativt låg; detta, tillsammans med framsteg inom andra områden såsom felrättande koder och sammanflätad synkronisering och avkodning, motiverar den fokus som tillägnas ickekoherenta metoder i avhandlingen. Det andra fokuseringsområdet är länkadaption, samt de mekanismer som krävs för dess implementation i adaptiva undervattensnätverk.

Avhandlingen ger en bakgrundsbeskrivning av olika angreppssätt inom akustisk signalering, både koherenta och ickekoherenta, samt andra nyckelkomponenter i en ickekoherent undervattenslänk, såsom symbolalfabet, datamodell hos motagaren, felrättande koder, samt synkronisering av tid och Doppler. Den bidrar också med diskussioner kring länkadaption i akustiska undervattensnätverk och forskningsmetodologi inom undervattenskommunikation. De inkluderade artiklarna tillhandahåller ett antal länkmetoder vars spektraleffektivitet är mellan 0.02-0.22 (bit/s/Hz), med olika krav på signal-brusförållande och längd på kommunikationsramen, som möjliggörs tack vare de presenterade framstegen inom ickekoherenta länkmetoder. Robustheten hos en mjuk avkodare som använder en Rice-fädande modell förbättras markant genom att begränsa signal-brusförhållandet hos modellens parametrisering; detta resultat är applicerbart på alla metoder som använder digital frekvensmodulation (FSK). Ett nytt ickekoherent symbolalfabet med upp till 1 (bit/s/Hz) spektraleffektivitet presenteras, vars dimensionalitet $M$ ökar prestandan hos en mjuk avkodare, och som är ekvivalent med så kallad on-off keying (OOK) för M=1. Ett ramverk för sammanflätad synkronisering och avkodning föreslås, som påvisats i en svår horisontell kanal med platformar vars relativa hastighet uppgick till +/- 4 (m/s). Dessutom presenteras ett ramverk för effektiv utvärdering av länkadaptionsalgoritmer, samt ett ad-hoc länkadapterande undervattensnätverk som använder implicit återkoppling med låg fördröjning, som också påvisats genom sjöförsök.

Forskningen som presenteras här har utförts som en del av Swedish Maritime Robotics Centre (SMaRC), ett nationellt tvärvetenskapligt forskningscenter finansierat av Stiftelsen för Strategisk Forskning (SSF). 

To make full use of the underwater acoustic channel, a high degree of adaptivity is typically required. A novel framework for efficient evaluation of link adaptation employing channel replay simulation with Watermark and a lookup table is proposed. The approach is used to compare link adaptation based on SNR and CRC feedback using measured sea data. The best performance is obtained with a CRC criterion since it detects if a communication link has poor behaviour. The method is further improved by introducing a memory mechanism that disables a link with repeated low reliability.

This paper investigates the implications of time variability in the underwater acoustic channel on coded multiple tone frequency-shift keying (FSK) that uses a Rice model on narrow toneband envelopes. It is evaluated with Convolution coded frequency parallel binary FSK in the 5 Watermark replay channels, in terms of the bit error rate over the noise measure E-b/N-0, where the model parameters are tracked using known pilots. The performance varies with pilot sparseness; however, it is significantly improved by limiting the likelihood input signal-to-noise ratio, which enables using a simple data model with minimal estimation overhead, and pilots can be separated by up to 10 [s].

Two MFSK modulation methods with time-frequency guards, using an LLR-based List Polar Decoder, are investigated for application to the underwater acoustic channel. Based on a number of published channel delay spreads, a 32FSK and a 32 subband 4FSK are chosen, having 185- and 1080 [bit/s], which are simulated in an AWGN channel, and tested at two sites in the Stockholm archipelago. The Polar Code is shown to have good performance in terms of BER and BLER, for both long and short frame lengths. 

With regards to energy constrained Autonomous Underwater Vehicles (AUVs), and difficulties inherent to the acoustic underwater communication channel, a non-coherent method is investigated in order to improve energy consumption and reliability over traditional Frequency Shift Keying (FSK), without increasing the bandwidth. A proposed method of adapting Trellis Coded Modulation (TCM) to constant amplitude Permutated Frequency Shift Keying (PFSK) symbol constellations is evaluated. A system implementation of two PFSK methods is simulated in an Additive White Gaussian Noise channel, and field tested in an underwater channel in the Stockholm archipelago, where a binary FSK reference method is used as a comparison. The main interest is comparing electrical bit energy and bit error rate (BER) for the methods. Time variability of frequency fading, related to wind speed, is also evaluated from the field tests. 

The underwater acoustic environment is known for its unpredictability, making it challenging to establish configuration parameters for acoustic modems before network deployment. When the modems are configured for robustness, potential throughput is often sacrificed; meanwhile, opting for high-rate links can result in communication failures in highly dynamic acoustic conditions. Given these challenges, this paper presents an adaptation framework for networked underwater acoustic modems. Its primary objective is to let modems adaptively select communication links that balance information rate and reliability. It is assumed that the modems provide a set of pre-configured links with monotonically increasing information rate and decreasing reliability. The framework is developed specifically for flooding-based routing protocols, which efficiently handle sudden changes in network topology. By leveraging existing network traffic and implicit acknowledgments, the framework achieves link adaptation with minimal network overhead, necessitating only the addition of a "previous node" address field in the packet headers. Field experiments conducted in a time-varying acoustic environment, using modems configured with four different links, show an increase in the average information per packet by a factor of up to 12, and a reduction in network transmission time of 25\%--50\%, demonstrating DLink's ability to enhance channel utilization significantly, outperforming configurations that rely solely on robust links. This improvement indicates DLink's potential to substantially increase the throughput of underwater acoustic networks.