Design
Data
SETI@home's data is recorded at 5 Mbps at the Arecibo Observatory in Puerto Rico. SETI@home used 35 GB digital linear tapes. The recording time per tape is 16 hours.
With one-bit complex sampling this yields a frequency band of 2.5 MHz which is enough to handle doppler shifts for relative velocities of up to 260 km/sec (or about the rate of the Milky Ways galactic rotation).
The frequency-band is like many other SETI-projects centered at the Hydrogen-line (1.42 GHz) because man-made transmissions are forbidden here by an international treaty.
Computational Model
SETI@home's computational model is simple. The signal data is divided into fixed-size work units distributed via the Internet to the clients. The client program computes a result (a set of candidate signals) and returns it to the server. As mentioned in the introduction there is no communication between clients.
SETI@home employs redundant computation. Each work unit is processed multiple times to compensate the detection and discard of results of faulty processors or malicious users. A redundancy level of 2 or 3 is adequate for this.
Work Units
The generation of work units happens at a bounded rate. Also clients asking for work are never turned away. This increases the redundancy level the more users the project has and the higher the average speed per user gets. As the quantities increased greatly over time the clients have been revised to do more computation per work unit in order to keep the level of redundancy at the intended level. The server complex creates and distributes the work units to minimizes tape handling.
Work units are formed by dividing the 2.5 MHz signal into 256 frequency bands. Each band is then divided into 107-second segments overlapping in time by 20 seconds. As SETI@home looks for continues signals with a length of up to 20 seconds the overlapp ensures that each signal is contained in at least one work unit. The resulting work units are 350 KB. Enough to keep a client busy for a while and small enough to be transfered in a matter of minutes even by a 56K Modem.
Server
The server complex contains a relational database which stores information about tapes, work units, results, users and other aspects of the project. A multi-threaded data/result server handles distribution of work units to clients. HTTP is used so that users can participate in SETI@home even when connecting from behind a firewall. A garbage collector program removes work units from disk clearing an on-disk flag in their database records. Work units are flaged depending on how many correct results were received from the clients versus the intended level of redundancy.
Computational work of the clients
Analyzing work units involves computing signal power as a function of frequency and time then looking for patterns:
- "spikes" (short bursts)
- "Gaussians" (narrow-bandwidth signals with a 20-second Gaussian envelope corresponding to the telescope’s beam movement across a point)
-"pulsed signals" (Gaussian signals pulsed with arbitrary period, phase, and duty cycle)
-"triplets" (three equally spaced spikes at the same frequency, or a simple pulsed signal)
A good german site for more details is setigermany.de.
Signals whose power and goodness of fit exceed thresholds are recorded in the output file.
--Ertelt 15:30, 31 Oct 2005 (CET)