Abstract-Inspired by the compute-and-forward scheme from Nazer and Gastpar, a novel multiple-access scheme introduced by Zhu and Gastpar makes use of nested lattice codes and sequential decoding of linear combinations of codewords to recover the individual messages. This strategy, coined computeforward multiple access (CFMA), provably achieves points on the dominant face of the multiple-access capacity region while circumventing the need of time sharing or rate splitting. For a two-user multiple-access channel (MAC), we propose a practical procedure to design suitable codes from off-the-shelf LDPC codes and present a sequential belief propagation decoder with complexity comparable with that of point-to-point decoders. We demonstrate the potential of our strategy by comparing several numerical evaluations with theoretical limits.

We experimentally achieve a 19% capacity gain per Watt of electrical supply power in a 12-span link by eliminating gain flattening filters and optimizing launch powers using machine learning by deep neural networks in a massively parallel fiber context. Gradient Descent DNN Launch Power Allocation Received Signal, Noise Powers

No abstract

Wyner’s common information is a measure that quantifies and assesses the commonality between two random variables. Based on this, we introduce a novel two-step procedure to construct features from data, referred to as Common Information Components Analysis (CICA). The first step can be interpreted as an extraction of Wyner’s common information. The second step is a form of back-projection of the common information onto the original variables, leading to the extracted features. A free parameter γ controls the complexity of the extracted features. We establish that, in the case of Gaussian statistics, CICA precisely reduces to Canonical Correlation Analysis (CCA), where the parameter γ determines the number of CCA components that are extracted. In this sense, we establish a novel rigorous connection between information measures and CCA, and CICA is a strict generalization of the latter. It is shown that CICA has several desirable features, including a natural extension to beyond just two data sets.

No abstract

In the problem of coded caching for media delivery, two separate coding opportunities have been identified. The first opportunity is a multi-user advantage and crucially hinges on a public broadcast link in the delivery phase. This has been explored in a plethora of works. The second opportunity has received far less attention and concerns similarities between files in the database. Here, the paradigm is to cache "the similarity" between the files. Upon the request, the encoder refines this by providing the specific details for the requested files. Extending Gray and Wyner's work (1974), it follows that the right measure of file similarity is Wyner's Common Information and its generalizations. The present paper surveys and extends the role of Wyner's Common Information in caching. As a novel result, explicit solutions are found for the Gaussian case under mean-squared error, both for the caching problem as well as for the network considered by Gray and Wyner. Our solution leverages and extends the recent technique of factorization of convex envelopes.

Inspired by the compute-and-forward scheme from Nazer and Gastpar, a novel multiple-access scheme introduced by Zhu and Gastpar makes use of nested lattice codes and sequential decoding of linear combinations of codewords to recover the individual messages. This strategy, coined computeforward multiple access (CFMA), provably achieves points on the dominant face of the multiple-access capacity region while circumventing the need of time sharing or rate splitting. For a two-user multiple-access channel (MAC), we propose a practical procedure to design suitable codes from off-the-shelf LDPC codes and present a sequential belief propagation decoder with complexity comparable with that of point-to-point decoders. We demonstrate the potential of our strategy by comparing several numerical evaluations with theoretical limits.

We experimentally achieve a 19% capacity gain per Watt of electrical supply power in a 12-span link by eliminating gain flattening filters and optimizing launch powers using deep neural networks in a parallel fiber context.