Nonlinear modulation is a promising technique for ultrasonic non-destructive damage identification. A wireless sensor network is ideally suited to monitor large structures using nonlinear modulation in a cost-efficient manner. However, existing approaches rely on high sampling rates and resource-demanding computations that are not feasible on low-cost and low-power sensor network devices. We present a new damage indicator that uses the short-time Fourier transform to derive amplitude and phase modulation with less computational effort and memory usage. Evaluation of the proposed method using real experiment data exhibits performance and reliability similar to the conventionally used modulation index. Undersampling is demonstrated, which reduces the memory demand in a test scenario by more than 100 times, and the required energy for sampling and processing more than four times. The loss of accuracy introduced by undersampling is shown to be negligible.

Sensors for industrial and structural health monitoring are often in shielded and hard-to-reach places. Acoustic wireless power transfer (WPT) and piezoelectric backscatter enable batteryless sensors in such scenarios. Although the low efficiency of WPT demands power-conserving sensor nodes, backscatter communication, which consumes near-zero power, has not yet been combined with WPT. This study reviews the available approaches to acoustic WPT and active and passive acoustic through-metal communication. We design a batteryless and backscattering tag prototype from commercially available components. Analysis of the prototypes reveals that low-power hardware poses additional challenges for communication, i.e., unstable and inaccurate oscillators. Therefore, we implement a software-defined receiver using digital phase-locked loops (DPLLs) to mitigate the effects of oscillator instability. We show that DPLLs enable reliable backscatter communication with inaccurate clocks using simulation and real-world measurements. Our prototype achieves communication at 2 kBs${}^{-1}$ over a distance of 3 m. Furthermore, during transmission, the prototype consumes less than 300 $\mathsf{\mu }$W power. At the same time, over 4 mW of power is received through wireless transmission over a distance of 3 m with an efficiency of 2.8%.

Receiver implementations with very low quantization resolution will play an important role in 5G, as high precision quantization and signal processing are costly in terms of computational resources and chip area. Therefore, low resolution receivers with quasi optimum performance will be required to meet complexity and latency constraints. The Information Bottleneck method allows for a novel, information centric approach to design such receivers. The method was originally introduced by Naftali Tishby et al. and mostly used in the machine learning field so far. Interestingly, it can also be applied to build surprisingly good digital communication receivers which work fundamentally different than state-of-the-art receivers. Instead of minimizing the quantization error, receiver components with maximum preservation of relevant information for a given bit width can be designed. All signal processing in the resulting receivers is performed using only simple lookup operations. In this paper, we first provide a brief introduction to the design of receiver components with the Information Bottleneck method. We keep referring to decoding of low-density parity-check codes as a practical example. The focus of the paper lies on practical decoder implementations on a digital signal processor which illustrate the potential of the proposed technique. An Information Bottleneck decoder with 4 bit message passing decoding is found to outperform 8 bit implementations of the well-known min-sum decoder in terms of bit error rate and to perform extremely close to an 8 bit belief propagation decoder, while offering considerably higher net decoding throughput than both conventional decoders.

We present an information-theoretic approach to discrete signal processing called information-optimum signal processing for the example of a wireless receiver comprising LDPC decoding, channel estimation and detection. All operations are replaced by simple lookup tables and all messages which are exchanged between detection stages are unsigned integers. The lookup tables are designed offline using the information bottleneck concept of preserving relevant information. We show that the approach allows for simple signal processing with coarse quantization while achieving virtually the same performance as conventional signal processing approaches with high resolution, e.g. double precision. The contribution of the paper is a tutorial style explanation of the concept as well as an exemplary survey of applications and performance results which illustrate the potential of the concept.