sd-wavys
distributed ocean monitoring system
source code: https://github.com/joaoesteves10/wavys
Context
Distributed Systems coursework, built in two phases: it simulates a network of ocean buoys (WAVY) that collect sensor data, send it to an AGGREGATOR that pre-processes it, which then forwards the data to a central SERVER that persists and analyses it. The first phase used point-to-point TCP sockets; the second evolved the communication between the WAVY and the AGGREGATOR into publish/subscribe over RabbitMQ (one topic per sensor) and introduced two independent gRPC microservices, for pre-processing and statistical analysis, while keeping the first phase's message format and ACK/NACK pattern to preserve compatibility. Besides the console interface, the SERVER exposes a REST API consumed by a Blazor web dashboard that lets you filter the received data by sensor, WAVY or time range, request new analyses, and export the filtered results as JSON.
Technical challenges
- pub/sub and RPC side by side — RabbitMQ for asynchronous, decoupled ingestion of WAVY data (one topic per sensor); gRPC/Protobuf for the synchronous pre-processing and analysis calls, each paradigm used where its semantics fit best
- missing data — resampling at fixed intervals with linear interpolation to estimate values at timestamps with no reading, before sending to the server
- configurable pre-processing — each sensor can request a different output format (float, json, csv, xml) from the pre-processing microservice, set via configuration file
- compatibility across phases — switching from sockets to RabbitMQ kept the first phase's plain text message format and ACK/NACK confirmations
- web dashboard with filtered export — a Blazor dashboard that consumes the server's REST API to filter data by sensor, WAVY or time range, request analyses (average, outliers, spikes, trend) and download the filtered result as JSON