Log-FSK: A Frequency Modulation for Over-the-Air Computing

In this study we introduce Logarithmic Frequency Shift Keying (Log-FSK), a novel frequency modulation for over-the-air computing (AirComp). Log-FSK leverages non-linear signal processing to produce AirComp in the frequency domain, this is, the maximum frequency of the received signal corresponds to the sum of the individual transmitted frequencies. The demodulation procedure relies on the inverse Discrete Cosine Transform (DCT) and the extraction of the maximum frequency component. We provide the theoretical performance in terms of error probability and mean squared error. To demonstrate its practicality, we present specific applications and experimental results showcasing the effectiveness of Log-FSK AirComp within linear Wireless Sensor Networks (WSN). Our experiments show that Log-FSK outperforms linear AirComp, implemented with double sideband (DSB), when working above the threshold SNR.

Adaptive function approximation based on the Discrete Cosine Transform (DCT)

This paper studies the cosine as basis function for the approximation of univariate and continuous functions without memory. This work studies a supervised learning to obtain the approximation coefficients, instead of using the Discrete Cosine Transform (DCT). Due to the finite dynamics and orthogonality of the cosine basis functions, simple gradient algorithms, such as the Normalized Least Mean Squares (NLMS), can benefit from it and present a controlled and predictable convergence time and error misadjustment. Due to its simplicity, the proposed technique ranks as the best in terms of learning quality versus complexity, and it is presented as an attractive technique to be used in more complex supervised learning systems. Simulations illustrate the performance of the approach. This paper celebrates the 50th anniversary of the publication of the DCT by Nasir Ahmed in 1973.

ENN: A Neural Network with DCT-Adaptive Activation Functions

The expressiveness of neural networks highly depends on the nature of the activation function, although these are usually assumed predefined and fixed during the training stage. In this paper we present Expressive Neural Network (ENN), a novel architecture in which the non-linear activation functions are modeled using the Discrete Cosine Transform (DCT) and adapted using backpropagation during training. This parametrization keeps the number of trainable parameters low, is appropriate for gradient-based schemes, and adapts to different learning tasks. This is the first non-linear model for activation functions that relies on a signal processing perspective, providing high flexibility and expressiveness to the network. We contribute with insights in the explainability of the network at convergence by recovering the concept of bump, this is, the response of each activation function in the output space to provide insights. Finally, through exhaustive experiments we show that the model can adapt to classification and regression tasks. The performance of ENN outperforms state of the art benchmarks, providing up to a 40\% gap in accuracy in some scenarios.

Frequency Modulation Aggregation for Federated Learning

Federated edge learning (FEEL) is a framework for training models in a distributed fashion using edge devices and a server that coordinates the learning process. In FEEL, edge devices periodically transmit model parameters to the server, which aggregates them to generate a global model. To reduce the burden of transmitting high-dimensional data by many edge devices, a broadband analog transmission scheme has been proposed. The devices transmit the parameters concurrently using a linear analog modulation, which are aggregated by the superposition nature of the wireless medium. However, linear analog modulations incur in an excessive power consumption for edge devices and are not suitable for current digital wireless systems. To overcome this issue, in this paper we propose a digital frequency broadband aggregation. The scheme integrates a Multiple Frequency Shift Keying (MFSK) at the transmitters and a type-based multiple access (TBMA) at the receiver. Using concurrent transmission, the server can recover the type (i.e., a histogram) of the transmitted parameters and compute any aggregation function to generate a shared global model. We provide a extensive analysis of the communication scheme in an AWGN channel and compare it with linear analog modulations. Our experimental results show that the proposed scheme achieves the same performance, although it requires 14 dB less in peak-to-average power ratio (PAPR) than linear analog modulations.

LoRa-based Over-the-Air Computing for Sat-IoT

Satellite Internet of Things (Sat-IoT) is a novel framework in which satellites integrate sensing, communication and computing capabilities to carry out task-oriented communications. In this paper we propose to use the Long Range (LoRa) modulation for the purpose of estimation in a Sat-IoT scenario. Then we realize that the collisions generated by LoRa can be harnessed in an Over-the-Air Computing (AirComp) framework. Specifically, we propose to use LoRa for Type-based Multiple Access (TBMA), a semantic-aware scheme in which communication resources are assigned to different parameters, not users. Our experimental results show that LoRa-TBMA is suitable as a massive access scheme, provides large gains in terms of mean squared error (MSE) and saves scarce satellite communication resources (i.e., power, latency and bandwidth) with respect to orthogonal multiple access schemes. We also analyze the satellite scenarios that could take advantage of the LoRa-TBMA scheme. In summary, that angular modulations, which are very useful in satellite communications, can also benefit from AirComp.

DCT-based Air Interface Design for Function Computation

With the integration of communication and computing, it is expected that part of the computing is transferred to the transmitter side. In this paper we address the general problem of Frequency Modulation (FM) for function approximation through a communication channel. We exploit the benefits of the Discrete Cosine Transform (DCT) to approximate the function and design the waveform. In front of other approximation schemes, the DCT uses basis of controlled dynamic, which is a desirable property for a practical implementation. Furthermore, the proposed modulation allows to recover both the measurement and the function in a single transmission. Our experiments show that this scheme outperforms the double side-band (DSB) modulation in terms of mean squared error (MSE). This can also be implemented with an agnostic receiver, in which the function is unknown to the receiver. Finally, the proposed modulation is compatible with some of the existing transmission technologies for sensor networks.