The most sensitive SETI observation by multi-beam coincidence matching strategy towards exoplanet systems

‘Are we alone?’ The Search for Extra-Terrestrial Intelligence (SETI) aims to answer this 14 profound question. Apart from examining environments in our solar system and detecting biosignatures in exoplanet atmospheres 1 SETI is another main approach to search for life be- yond Earth by detecting technosignatures indicative of extra-terrestrial intelligence (ETI) 2,3 , 17 such as engineering radio signals. Massive e ﬀ orts have been made by SETI scientists using some none as an ETI technosignature so far. Most targeted SETI observa- tions performed in recent years use on-o ﬀ strategy to distinguish signals transmitted from celestial bodies from radio frequency interference (RFI) generated near the ground. Here we report a SETI campaign employing another SETI observation strategy, multi-beam coin- cidence matching (MBCM), at the Five-hundred-meter Aperture Spherical radio Telescope 24 (FAST) towards 33 currently discovered planetary systems, searching for narrow band drift- ing signals at a band of 1 . 05 − 1 . 45 GHz. Our observations achieve an unprecedented sensi- tivity with a minimum Equivalent Isotropic Radiated Power (EIRP) of 1 . 5 × 10 9 W. We pro- cess the data of two orthogonal polarization separately, aiming to discriminate instrumental RFI signals. A particular signal at 1140.604 MHz from the observation towards Kepler-438 passes our initial selection criteria. Although we have not yet determined the exact cause of this signal, its polarization characteristic suggests that it is most likely to be attributed to RFI. In spite of this, our work veriﬁes that compared to single-beam on-o ﬀ strategy, MBCM greatly improves both time e ﬃ ciency and radio frequency interference (RFI) identiﬁcation ﬀ ectiveness of targeted SETI observations.

1 radio telescopes around the world 4-17 . Though some candidate signals were detected, none 19 of them has been confirmed as an ETI technosignature so far. Most targeted SETI observa-20 tions performed in recent years use on-off strategy to distinguish signals transmitted from 21 celestial bodies from radio frequency interference (RFI) generated near the ground. Here 22 we report a SETI campaign employing another SETI observation strategy, multi-beam coin-23 cidence matching (MBCM), at the Five-hundred-meter Aperture Spherical radio Telescope 24 (FAST) towards 33 currently discovered planetary systems, searching for narrow band drift-25 ing signals at a band of 1.05 − 1.45 GHz. Our observations achieve an unprecedented sensi-26 tivity with a minimum Equivalent Isotropic Radiated Power (EIRP) of 1.5 × 10 9 W. We pro-27 cess the data of two orthogonal polarization separately, aiming to discriminate instrumental 28 RFI signals. A particular signal at 1140.604 MHz from the observation towards Kepler-438 29 passes our initial selection criteria. Although we have not yet determined the exact cause 30 of this signal, its polarization characteristic suggests that it is most likely to be attributed to 31 RFI. In spite of this, our work verifies that compared to single-beam on-off strategy, MBCM 32 greatly improves both time efficiency and radio frequency interference (RFI) identification 33 effectiveness of targeted SETI observations. 34 With an enormous collecting area (illuminated aperture of 300 m), a large sky region cov-35 erage (−14 • to +66 • in declination) and the cryogenically-cooled L-band 19-beam receiver (sys-36 tem temperature of ∼20 K), FAST 18-20 is well positioned to conduct highly sensitive and efficient 37 searches for ETI technosignatures 21 , and SETI is one of the five key science goals specified in 38 the original FAST project plan 22 . In 2019, the first commensal SETI survey by FAST was per-39 2 formed and two groups of candidate signals were detected 23 . Here in this work, we present the 40 first targeted SETI observations by FAST. From 2020 November to 2021 September, 33 currently 41 discovered planetary systems have been observed (Fig. 5), including 29 systems hosting planets in 42 their habitable zones [24][25][26] and 5 systems in the Earth transit zone 27 , namely, worlds that resemble 43 ours and worlds that can see us. 44 The greatest challenge for SETI observations is RFI identification and excision 23 . In re-45 cent years, targeted SETI observations adopt on-off strategy for this purpose 10, 13-17 , alternating 46 the telescope pointing between a target (on-source) and several reference locations (off-source).

47
The angular distance between an on-source and an off-source should be at least several times the 48 full width at half maximum (FWHM), of the telescope, thus ensuring signals transmitted from the 49 on-source cannot be detected from the off-source. Ubiquitous RFI can enter the side lobes of the 50 beam, so it can be detected in both on-and off-observations. Hence, signals detected from both 51 the on-source and any of the off-sources are identified as RFI and are removed directly, while the 52 remaining filtered signals will be further examined.

53
However, on-off strategy is inefficient for FAST because of its lengthy slewing time, which 54 costs ∼10 min per pointing 20 . Thanks to the FAST L-band 19-beam receiver 18-20 , the observation 55 efficiency can be improved significantly by multi-beam observation. Based on the principles of 56 on-off strategy, we design the multi-beam coincidence matching strategy for targeted SETI obser-57 vations by FAST. During an observation, all 19 beams on the receiver record data simultaneously 58 while beam 1 points to the target. The angular distance between two adjacent beams is ≲ 6 ′ , about 59 3 twice the L-band FWHM (2.9 ′ ). In addition to beam 1, signals transmitted from the target are 60 likely to cover several of the 6 beams adjacent to beam 1, but are impossible to cover the 6 out-61 ermost beams (beam 8, 10, 12, 14, 16 and 18), which are ∼4 FWHM from beam 1. Therefore, 62 these 6 beams serve as reference locations like the off-sources used in on-off strategy, and signals 63 detected by both beam 1 and any of these 6 beams are rejected directly (Fig.1a). Using as many as 64 6 reference beams can effectively reduce the contingency caused by low S/N and signal incident 65 direction.

66
The preliminary screening of signals only uses the data of these 7 beams, but can remove the 67 vast majority of signals detected by beam 1. Then we require the data of other 12 beams to check 68 the beam coverage of the remaining signals. We stipulate that an extra-terrestrial signal cannot 69 cover two beams that are separated by one or more beams, otherwise it is RFI. This criterion can 70 also be used to conduct blind search in the sky region around observed targets, making the most of  On-off strategy cannot identify RFI signals that happen to be generated when observing on-source 77 and interrupted when observing off-source, such as signals with a duty cycle matching the on-  Doppler effect, and the drift rate is given by where ν 0 is the emitted frequency from the transmitter, and a is the relative acceleration between the 91 transmitter and the receiver. Because the relative acceleration between the Earth and an exoplanet 92 is unlikely to be exact zero, non-drift signals are determined as RFI directly, which originate from 93 stationary ground-based interference sources and make up a considerable proportion of detected 94 signals.

95
Some previous targeted SETI searched solely Stokes I data for narrow band drifting signals 14,17 .

96
The data recorded by FAST spectral line backend consists of four polarizations, among which po-97 larization 1 and 2 (pol1 and pol2) represent the intensity of two orthogonal directions (I X and I Y ) 98 respectively. We process the data of two orthogonal polarizations separately because we find that 99 instrumental RFI usually appears much stronger in one polarization than another, such as harmon-100 ics produced by crystal oscillators, a major type of instrumental RFI. This imbalance is caused by 101 the relative positions of the instrumental RFI sources and the data pipelines of two polarizations, 102 and the different shielding levels of the pipelines to instrumental RFI. A signal that appears much 103 stronger in one polarization than another is unlikely to come from an extra-terrestrial source. Even 104 though ETI may emit linearly polarized signals, this phenomenon can hardly occur unless the po-105 larization direction of the signal is almost aligned with one of the two polarization directions of 106 the receiver during the observation and is not seriously affected by interstellar polarization.

107
Using turboSETI 29 , a software package for targeted SETI, we search our data for narrow 108 6 band drifting signals across a frequency range of 1.05 − 1.45 GHz (FAST L-band) 18, 20, 21 . The 109 narrow band signals detected above the signal-to-noise ratio (S/N) threshold (taken as 10 in this 110 work) are referred to as 'hits' [13][14][15]17 , and hits that detected only by the target beam (beam 1) but 111 not by any of the reference beams (beam 8, 10, 12, 14, 16 and 18) are referred to as 'events'. We 112 find 1,309,503 hits from pol1 and 1,324,198 hits from pol2, among which we select 2,013 events 113 from pol1 and 2,064 events from pol2, excluding 97.0% and 96.9% of the hits detected by beam 1 114 respectively.

115
The distributions of frequency, drift rate and S/N for detected hits and events are shown in terfall plots) and detailed analysis of frequency and drift rate. All the events that have been con-129 firmed as RFI can be classified into two categories. The first type of events are false positives, 130 including two cases. One case is that there is no hit present on the waterfall plot of beam 1, but 131 turboSETI detects a hit, while the other case is that there are hits present on the waterfall plots of 132 some of the reference beams, but turboSETI fails to detect them. The second type of events are in-133 strumental RFI resulting from inner interference sources, specifically, crystal oscillator harmonics. we detect. According to this feature and the fact that this event persists for 20 min, during which 139 its drift rate varies slightly, we can exclude the possibility of all ground-based RFI sources outside 140 the telescope, including airplanes. We also find that no satellite or deep-space probe entered the 141 main lobe of beam 1 during the observation, thus the possibility of artificial objects is also ruled 142 out.

143
Although some of its characteristics are somewhat consistent with a true ETI signal, there 144 is still a piece of evidence leading us to suspect that the Kepler-438 event is an instrumental RFI 145 signal. This event appears much stronger in pol2 than pol1, as shown in Fig.6. This is consistent 146 with the aforementioned property of crystal oscillator harmonics. But its frequency has no relation 147 with the fundamental frequencies of the crystal oscillators in the FAST instruments, and it only 148 appears in beam 1, thus it is unlikely to be associated with crystal oscillator harmonics. So far, we 149 are still uncertain about the exact origin and generation mechanism of the Kepler-438 event, and 150 more experimental re-observations are required. Even if this event is determined as instrumental 151 RFI, the results will provide meaningful experience for RFI identification in future SETI.

152
The sensitivity of radio SETI observations can be measured by equivalent isotropic radiated 153 power(EIRP) 13,14,17,21 , defined as where d is the distance to the target, σ is the S/N threshold, T sys is the system temperature, A eff   simultaneously, an RFI signal is expected to appear at the same frequency in all beams. We modify 272 the code and set the RFI rejection range as ±3δν, where δν is our frequency resolution. That is, if 273 a hit present in beam 1 is accompanied by any hit within this range in the reference beams, the hit 274 will be determined as RFI.

275
Signal identification As mentioned above, we can rank SETI signal from low to high level as: 276 hit, event, candidate and signal of interest or ETI signal. Events passing the find event pipeline 277 of turboSETI are not certainly real candidate signals, and every event requires examination from 278 many aspects. another.

301
In this work, we can exclude all the events selected by turboSETI except the Kepler-438 302 event through the above procedures. The following procedures are used to further examine this 303 particular event and applicable to various events that cannot be excluded by the above procedures.  none of them has the same morphology with the Kepler-438 event.

335
For the latter case, the beam coverage should be compared between the particular event and 336 21 possible homologous RFI to determine whether they share the same origin. From observations to-337 wards other 8 sources, we find 8 weak drifting signals in the range of 1140.604 MHz ± 2.5 kHz. All 338 the 8 signals appear only in beam 1 and pol2, but fail to reach the S/N threshold (Fig.7). However, June 29. Hence, we infer that these 8 signals are correlated and generated by the same RFI source, 343 but the Kepler-438 event is unlikely to come from the same source as that of these 8 signals.   show the plots of pol1 and the right panels show the plots of pol2. The frequencies of the latter 8 signals (red lines) are lower than that of the first signal (yellow lines).