The traditional AE systems manipulate the signal in analog domain. AE events are detected and important signal parameters like rise time, peak amplitude, energy, etc., are determined in analog domain. These parameters are transferred to digital domain for the computer processing. Even though analog method is easier for high frequency signals, it has several disadvantages. The hardware design is complex and in-flexible, and the system is affected by ageing, drifting, etc. Moreover, it is difficult to obtain the spectrum and related parameters in analog domain. The advent of high speed analog to digital converters (flash ADCs) and very powerful digital signal processors (DSP) made AE signal analysis viable in digital domain.

AEDAPS uses an array of DSPs to capture the AE data. Due to the high frequency content of the signal, the throughput of the incoming data stream is substantial (several million words per channel per second). In order to reduce the total data flow into the system, AEDAPS implements a threshold comparison logic for each channel. The threshold comparison level is software selectable. It acts as a gate that opens when the signal from the channel crosses it. Concurrent with this, the DMA logic for data transfer is enabled and a hardware interrupt is generated. On receipt of this, the DSP monitors the characteristics of the incoming signal. The DSP terminates the DMA transfer and closes the gate when the event is complete. This prevents further inflow of data into the system freeing the DSP to attend to the task of on-line processing of the acquired data. Each DSP element has a large pool of local memory to store the incoming data. This ensures that the data is not lost even in the case of a burst of events.

AEDAPS provides a flexible interconnection of the channels in a three layered hierarchical structure. At layer 1, any two channels can be brought to the same DSP. The DSPs themselves are interconnected in mesh topology using high speed data channels at layer 2.The channel assignment over this mesh is flexible and hence channels with high correlation can be assigned to neighbouring nodes. At the top most layer, host systems containing clusters of DSPs are interconnected over a bus. The AE data arriving at any channel may be made to pass through these computing nodes. Since the nodes are operating in parallel, the computational load could be effectively distributed facilitating detailed on-line processing.

Consistent with this hardware structure, the software for AEDAPS also is organised into three layers. The Data acquisition and pre-processing kernel (DAP) residing in the DSP boards computes the signal parameters. It can also pre-process the date before transmitting it to a neighbouring node in the mesh. The signal analysis and presentation software (SAP) in the host systems is graphic oriented and interactive. It consists of different functional modules for calibration, data acquisition, raw data presentation, data storage, source location identification, and joint time-frequency analysis. At the top level, an overall configuration manager co-ordinates the working of the DAP and SAP software modules. It takes care of channel assignments, processing steps at each node and data flow between nodes.

The major feature of AEDAPS over alternate designs is its flexibility. Depending on the channel and processing power requirements, the number of DSPs in the configuration may be progressively increased. This modularity is preserved in the software also. AEDAPS can provide valuable information about the emission events and its possible causes by the detailed on-line processing of single and cross channel AE data.