Heterogeneous embedded monitoring of common carp pond water quality

Abstract

Low-cost internet of things (IoT) systems are increasingly used in aquaculture, but many reports provide limited information on node architecture, signal processing, and validation scope. This paper presents a short-duration proof-of concept evaluation of a heterogeneous two-node embedded system for a common carp pond. Node 1 used a LattePanda as a sensing node, coordinator, sink, edge gateway, and cloud interface; Node 2 used an Arduino UNO and an XBee router. Dissolved oxygen was measured at Node 1, while pH and temperature were measured at both nodes. The deployment produced 396 complete cloud records over 2 h 5 min 43 s, with a median positive interval of 16 s. Dissolved oxygen averaged 5.363±0.105 mg/L. Against four field-reference readings, pH mean absolute errors (MAEs) were 0.133 and 0.078, and temperature MAEs were 0.175 and 0.525 ◦C for Nodes 1 and 2. The software calculated the U.S. Environmental Protection Agency chronic ammonia criterion from pH and temperature; it did not measure total ammonia nitrogen (TAN). The results demonstrate end-to-end acquisition, wireless aggregation, cloud storage, and mobile visualization while defining the limitations of short-duration and non-time-matched reference data.