Motivated by the hypothesis that 'Oumuamua could conceivably be an interstellar probe, we used the Allen Telescope Array to search for radio transmissions that would indicate a non-natural origin for this object. Observations were made at radio frequencies between 1 and 10 GHz using the Array's correlator receiver with a channel bandwidth of 100 kHz. In frequency regions not corrupted by man-made interference, we find no signal flux with frequency-dependent lower limits of 0.01 Jy at 1 GHz and 0.1 Jy at 7 GHz. For a putative isotropic transmitter on the object, these limits correspond to transmitter powers of 30 mW and 300 mW, respectively. In frequency ranges that are heavily utilized for satellite communications, our sensitivity to weak signals is badly impinged, but we can still place an upper limit of 10 W for a transmitter on the asteroid. For comparison and validation should a transmitter be discovered, contemporaneous measurements were made on the solar system asteroids 2017 UZ and 2017 WC with comparable sensitivities. Because they are closer to Earth, we place upper limits on transmitter power to be 0.1 and 0.001 times the limits for 'Oumuamua, respectively.

The Allen Telescope Array was used to search for signals with characteristics similar to the “Wow” signal, the best candidate for an extraterrestrial radio signal found during Ohio State University’s (OSU’s) seven-year 21 cm 10-kHz channel sky survey for signals possibly due to extraterrestrial intelligence. While previous follow-up searches have reported null results, our observations covered a 5 deg 2 field of view that extends well beyond the locus of all consistent directions of arrival (DOAs) of the original signal, and covered a 10 MHz bandwidth four times wider than the widest prior follow-up observations, using 12.8 kHz channels approximating OSU’s 10 kHz resolution. Approximately 100 hours of data were accumulated, considerably more time than any previous follow-up campaigns. We used interferometric imaging with an angular resolution of approximately 007 and automated feature-finding to search for point-like features mimicking a Wow repetition, obtaining single-channel sensitivity of ∼1.2 Jy for one minute averages. This allows identification of the DOA of a very brief repetition, with strong discrimination from radio interference, and eliminates the usual constraint that the signal must persist for long periods of time (around one hour) before the true DOA can be verified (because interfering signals from the horizon sometimes masquerade as coming from the look direction). No point-like features significantly exceeding the noise were found inside the full width at half maximum of the OSU fields of view, although one 26 σ point-like feature was detected during one 10 second integration about 1/3° away.

We apply classical machine vision and machine deep learning methods to prototype signal classifiers for the search for extraterrestrial intelligence. Our novel approach uses two-dimensional spectrograms of measured and simulated radio signals bearing the imprint of a technological origin. The studies are performed using archived narrow-band signal data captured from real-time SETI observations with the Allen Telescope Array and a set of digitally simulated signals designed to mimic real observed signals. By treating the 2D spectrogram as an image, we show that high quality parametric and non-parametric classifiers based on automated visual analysis can achieve high levels of discrimination and accuracy, as well as low false-positive rates. The (real) archived data were subjected to numerous feature-extraction algorithms based on the vertical and horizontal image moments and Huff transforms to simulate feature rotation. The most successful algorithm used a two-step process where the image was first filtered with a rotation, scale and shift-invariant affine transform followed by a simple correlation with a previously defined set of labeled prototype examples. The real data often contained multiple signals and signal ghosts, so we performed our non-parametric evaluation using a simpler and more controlled dataset produced by simulation of complex-valued voltage data with properties similar to the observed prototypes. The most successful non-parametric classifier employed a wide residual (convolutional) neural network based on pre-existing classifiers in current use for object detection in ordinary photographs. These results are relevant to a wide variety of research domains that already employ spectrogram analysis from time-domain astronomy to observations of earthquakes to animal vocalization analysis.

In this paper, we present our approach based on Convolutional neural networks for classification of narrow band signals. By converting the radio signals into 2D spectrogram image, the problem of signal classification can be transformed to as image classification problem. Deep convolutional networks are currently the state-of-art techniques in image classification. We demonstrate the effectiveness of CNN technique by using a simple Alexnet [2] network with 5 convolutional layers. Our approach achieved a score of 0.21 based on the logloss metric in the hackathon. In real world scenarios, telescopes can receive multiple types of narrow-band signals simultaneously. We also extend our work to detect simultaneously occurring signals with multiple labels and initial results for the same will be presented in this paper.

We present a photometric detection of the first brightness dips of the unique variable star KIC 8462852 since the end of the Kepler space mission in 2013 May. Our regular photometric surveillance started in October 2015, and a sequence of dipping began in 2017 May continuing on through the end of 2017, when the star was no longer visible from Earth. We distinguish four main 1-2.5 dips, named "Elsie," "Celeste," "Skara Brae," and "Angkor", which persist on timescales from several days to weeks. Our main results so far are: (i) there are no apparent changes of the stellar spectrum or polarization during the dips; (ii) the multiband photometry of the dips shows differential reddening favoring non-grey extinction. Therefore, our data are inconsistent with dip models that invoke optically thick material, but rather they are in-line with predictions for an occulter consisting primarily of ordinary dust, where much of the material must be optically thin with a size scale <<1um, and may also be consistent with models invoking variations intrinsic to the stellar photosphere. Notably, our data do not place constraints on the color of the longer-term "secular" dimming, which may be caused by independent processes, or probe different regimes of a single process.

We report a novel radio autocorrelation search for extraterrestrial intelligence. For selected frequencies across the terrestrial microwave window (1-10 GHz), observations were conducted at the Allen Telescope Array to identify artificial non-sinusoidal periodic signals with radio bandwidths greater than 4 Hz, which are capable of carrying substantial messages with symbol rates from 4 to 10(6) Hz. Out of 243 observations, about half (101) were directed toward sources with known continuum flux >similar to 1 Jy over the sampled bandwidth (quasars, pulsars, supernova remnants, and masers), based on the hypothesis that they might harbor heretofore undiscovered natural or artificial repetitive, phase or frequency modulation. The rest of the observations were directed mostly toward exoplanet stars with no previously discovered continuum flux. No signals attributable to extraterrestrial technology were found in this study. We conclude that the maximum probability that future observations like the ones described here will reveal repetitively modulated emissions is less than 5% for continuum sources and exoplanets alike. The paper concludes by describing a new approach to expanding this survey to many more targets and much greater sensitivity using archived data from interferometers all over the world.

Last year, the SETI Institute in collaboration with IBM, lead an effort to apply convolutional neural networks to the problem of identifying the class of unknown signals in simulated single-dish radio telescope data, with much success. In this paper we look at a different problem in radio astronomy, the construction of high fidelity images from stored correlator visibility files. This work is motivated by the amazing recent successes of CNN autoencoders developed for photographic image processing in the machine learning community. The second half of a photographic autoencoder has a structure very similar to what would be required for image generation from raw radio interferometer data (visibilities). In this paper, we argue that a deep convolutional neural network (CNN) can be a highly (computationally) effective approach to radio interferometer image generation starting from raw visibilities and producing high fidelity, cleaned (deconvolved) images as an output. We consider the linear and nonlinear operations performed in image generation and how they have analogs in a standard CNN. We also discuss the potential computational cost savings that might be had by replacing our complicated image processing pipelines with neural networks. A toy model CNN is developed and initial results will be presented.

We describe a new approach and algorithm for the detection of artificial signals and their classification in the search for extraterrestrial intelligence (SETI). The characteristics of radio signals observed during SETI research are often most apparent when those signals are represented as spectrograms. Additionally, many observed signals tend to share the same characteristics, allowing for sorting of the signals into different classes. For this work, complex-valued time-series data were simulated to produce a corpus of 140,000 signals from seven different signal classes. A wide residual neural network was then trained to classify these signal types using the gray-scale 2D spectrogram representation of those signals. An average $F_1$ score of 95.11\% was attained when tested on previously unobserved simulated signals. We also report on the performance of the model across a range of signal amplitudes.

We developed a new generation of low-noise, broadband feeds for the Allen Telescope Array at the Hat Creek Observatory in Northern California. The new feeds operate over the frequency range 0.9 to 14 GHz. The noise temperatures of the feeds have been substantially improved by cooling the entire feed structure as well as the low-noise amplifiers to 70 K. To achieve this improved performance, the new feeds are mounted in glass vacuum bottles with plastic lenses that maximize the microwave transmission through the bottles. Both the cooled feeds and their low-noise amplifiers produce total system temperatures that are in the range 25-30 K from 1 GHz to 5 GHz and 40-50 K up to 12.5 GHz.

We report on a search for the presence of signals from extraterrestrial intelligence in the direction of the star system KIC 8462852. Observations were made at radio frequencies between 1-10 GHz using the Allen Telescope Array. No narrowband radio signals were found at a level of 180-300 Jy in a 1 Hz channel, or medium band signals above 10 Jy in a 100 kHz channel.

We report radio SETI observations on a large number of known exoplanets and other nearby star systems using the Allen Telescope Array (ATA). Observations were made over about 19000 hours from May 2009 to Dec 2015. This search focused on narrow-band radio signals from a set totaling 9293 stars, including 2015 exoplanet stars and Kepler objects of interest and an additional 65 whose planets may be close to their Habitable Zone. The ATA observations were made using multiple synthesized beams and an anticoincidence filter to help identify terrestrial radio interference. Stars were observed over frequencies from 1- 9 GHz in multiple bands that avoid strong terrestrial communication frequencies. Data were processed in near-real time for narrow-band (0.7- 100 Hz) continuous and pulsed signals, with transmitter/receiver relative accelerations from -0.3 to 0.3 m/s^2. A total of 1.9 x 10^8 unique signals requiring immediate follow-up were detected in observations covering more than 8 x 10^6 star-MHz. We detected no persistent signals from extraterrestrial technology exceeding our frequency-dependent sensitivity threshold of 180 - 310 x 10^-26 W / m^2.

The Allen Telescope Array is a multi-user instrument and will perform simultaneous radio astronomy and radio SETI (search for extra-terrestrial intelligence) observations. It is a multi-beam instrument, with 16 independently steerable dual-polarization beams at 4 different tunings. Given 4 beams at one tuning, it is possible to distinguish RFI from true ETI signals by pointing the beams in different directions. Any signal that appears in more than one beam can be identified as RFI and ignored during SETI. We discuss the effectiveness of this approach for RFI rejection using realistic simulations of the fully populated 350 element configuration of the ATA as well as the interim 32 element configuration. Over a 5 minute integration period, we find RFI rejection ratios exceeding 50 dB over most of the sky.

As a new generation radio telescope, the Allen Telescope Array (ATA) is a prototype for the SKA. Here we describe recently developed design constraints for the ATA digital signal processing chain as a case study for SKA processing. As radio frequency interference (RFI) becomes increasingly problematical for radio astronomy, radio telescopes must support a wide range of RFI mitigation strategies including online deterministic and adaptive RFI nulling. We observe that at the ATA, the requirements for digital accuracy and control speed are not driven by astronomical imaging but by RFI. This can be understood from the fact that high precision is necessary to remove strong RFI signals from the weak astronomical background, and because RFI signals may change rapidly compared with celestial sources. We review and critique lines of reasoning that lead us to some of the design specifications for ATA digital processing. Introduction Some of the worst sources of radio frequency interference (RFI) for radio astronomy are low earth orbit (LEO) satellites. Unlike ground-based RFI sources, there is no place on earth where we can hide from such satellites. They broadcast strong signals that sometimes impinge on protected radio astronomical radio bands. Because of their low orbit, their angular position on the sky can vary faster than 1o per second. This presents a major challenge for astronomers attempting to reduce the damaging effects of these sources, and high speed calculations are required to simulate and remove their signals. The square kilometer array (SKA) must face up to this challenge. Because RFI mitigation is so important for the SKA’s success, it must be built into its design from the start. In this regard the Allen Telescope Array (ATA) provides a case study. Having just completed the detailed design, we report that ATA’s digital signal processing requirements are driven by RFI mitigation. Indeed, we discovered that at current technology and funding levels we cannot build a system that is flexible enough to support all desired methods of RFI suppression. In this paper we use simulations of active deterministic nulling of RFI from LEO satellites to quantify requirements for ATA signal processing. Although the simulations assume deterministic nulling, our results can be generalized to adaptive nulling and postcorrelation image processing. ATA Digital Processing and Simulations The ATA is a privately-funded interferometer currently under construction at Hat Creek Radio Observatory in northern California. It is being built in stages, first with 32 elements, then 206, then 350 elements for a total collecting area of about one hectare. The ATA data processing system has been developed and examined in several previous reports. , , , 3 4 5 6 Radio frequency signals from each antenna are downconverted and digitized with 150 MHz bandwidth. These signals are digitally delayed, downsampled to 100 MHz, and then fringe rotation is removed in the digital domain. After this processing, the signals are passed on to an imaging correlator , , 8 9 10 or to beamformers. The ATA beamformer is conceptually depicted in Fig. 1. This is a single-tap beamformer since only one value of delay, τ, is specified for each antenna signal before they are combined. After the delay, fringe rotation is corrected in each antenna signal with a single complex coefficient, c. RFI mitigation through synthetic beam pattern control is also accomplished via manipulation of the coefficients, c. This includes both deterministic and adaptive nulling. In this paper we focus on the requirements for digital control of τ and c. We discuss the communication interface between two subsystems of the ATA. The first is the ATA control software (host), which is distributed over multiple computers and linked by a local area network. The second subsystem is the IF processor, which consists of hundreds of custom-designed, field programmable gate array circuit boards. The host software precalculates τ and c and funnels this information to the IF processor where these coefficients are applied. The information content of these data is quite high, and it is desirable to find a representation that compresses τ and c to manageable rates. As we shall see, appropriate compression of this data is more subtle than it appears at first glance. To elucidate the information content of τ and c, we perform simulations with the currently proposed configuration of the ATA-350. For simplicity, the observation source is placed at telescope zenith. The RFI source is assumed to be in polar orbit on a path that passes directly over the observatory, and the simulations put the satellite in the vicinity of elevation 50o. These calculations use techniques we have developed specifically for the ATA, that generate wide frequency band nulls in a single-tap beamformer. We ignore antenna primary beam variations, whose angular scale is about 100 times larger than that of the synthetic beam pattern. This amounts to the assumption that the primary beam pattern is well characterized at each antenna element, and has been corrected for prior to synthetic beam formation. Specifying Delay, Phase and Amplitude We expect that RFI mitigation will be very important for the success of the ATA and design the data processing chain to be as flexible as possible for real-time RFI removal. This is especially important for devices that rely on beamformer outputs, since postcorrelation techniques for RFI removal are not possible in this case. With this in mind, we notice that some RFI sources are ground-stationary while others (satellites, airplanes) move across the sky at varying speeds. We begin with the assumption that RFI removal (e.g. deterministic or adaptive nulling) requires capabilities similar to tracking and forming a beam on the moving RFI source. At the ATA the requirement was set, somewhat arbitrarily, that we can track RFI sources moving at least as fast as a 350 km altitude low earth orbit (LEO) satellite. This includes almost all present or planned LEO’s but is not fast enough to follow the international space station or an airplane flying directly over the site. The linear velocity of a LEO is nearly independent of altitude and can be approximated by equating the centripetal force with the force of gravity, giving 8 ≈ ≈ rg v km/s. If we approximate the earth’s surface with a plane, we can estimate the angular velocity of a satellite (as viewed from the ground) in an overhead pass as a function of elevation angle θ , and altitude : h = = max 2 , sin θ θ θ & & h v 1.3o / s. [1] The ATA antennas are arrayed over a 1 km x 1 km area. For an antenna at the extreme edge of the array (d = 500 m from center), the path length difference relative to the array center is θ cos d p = , leading to a signal delay of θ θ τ cos 10 x 1.6 cos 6 − = = c d seconds. [2] From this we can estimate the maximal delay and maximal rate of change of delay . s / samples 6 10 x 4 ~ and , samples 000 1 s 10 x 7 4 ~

The systematic search for extraterrestrial intelligence (SETI) has been ongoing for slightly more than half the century that this journal has been in print, and the topic has been of human interest through recorded history. Are we any closer to detecting cosmic company or knowing whether it should be there? This paper takes a look at what has been done, what we are currently doing, and briefly speculates about what the future of SETI research may look like as well as the implications of a successful detection.

NASA’s LCROSS (Lunar Crater Observation and Sensing Satellite) mission was designed to explore the nature of previously detected enhanced levels of hydrogen near the lunar poles. The LCROSS mission impacted the spent upper stage of the launch vehicle into a permanently shadowed region of the lunar surface to create an ejecta plume. The resultant impact crater and plume were then observed by the LCROSS Shepherding Spacecraft as well as a cadre of telescopes on the Earth and in space to determine the nature of the materials contained within the permanently shadowed region. The Shepherding Spacecraft then became a second impactor which was also observed by multiple assets. The LCROSS Observation Campaign was a key component of the LCROSS mission. The goal of the Observation Campaign was to realize the scientific benefits of extending the LCROSS observations to multiple ground and space-based assets. This paper describes the LCROSS Observation Campaign and provides an overview of the Campaign coordination and logistics as well as a summary of the observation techniques utilized at a multitude of observatories. Lessons learned from the LCROSS Observation Campaign are also discussed to assist with the planning of future unique observing events.

In the cm-wavelength range, an extraterrestrial electromagnetic narrow band (sine wave) beacon is an excellent choice to get alien attention across interstellar distances because 1) it is not strongly affected by interstellar / interplanetary dispersion or scattering, and 2) searching for narrowband signals is computationally efficient (scales as Ns log(Ns) where Ns = number of voltage samples). Here we consider a special case wideband signal where two or more delayed copies of the same signal are transmitted over the same frequency and bandwidth, with the result that ISM dispersion and scattering cancel out during the detection stage. Such a signal is both a good beacon (easy to find) and carries arbitrarily large information rate (limited only by the atmospheric transparency to about 10 GHz). The discovery process uses an autocorrelation algorithm, and we outline a compute scheme where the beacon discovery search can be accomplished with only 2x the processing of a conventional sine wave search, and discuss signal to background response for sighting the beacon. Once the beacon is discovered, the focus turns to information extraction. Information extraction requires similar processing as for generic wideband signal searches, but since we have already identified the beacon, the efficiency of information extraction is negligible.

One of the exciting prospects for large N arrays is the potential for custom beam forming when operating in phased array mode. Pattern nulls may be generated by properly weighting the signals from all antennas with only minor degradation of gain in the main beam. Here we explore the limits of beam shape manipulation using the parameters of the Allen Telescope Array. To generate antenna weights, we apply an iterative method that is particularly easy to understand yet is comparable to linearly-constrained methods. In particular, this method elucidates how narrow band nulls may be extended to wider bandwidth. In practical RFI mitigation, the gain in the synthetic beam is obviously affected by the number and bandwidth of nulls placed elsewhere. Here we show how to predict the impact of a set of nulls in terms of the area of sky covered and null bandwidth. Most critical for design of the ATA, we find that high-speed (~10 ms) amplitude control of each array element over the full range 0-1 is critically important to allow testing of wide area / wide bandwidth nulling.

The Search for ExtraTerrestrial Intelligence (SETI) finally has its own full-time telescope. The Allen telescope array (ATA) in Northern California was dedicated on October 11, 2007. This array, which will eventually be composed of 350 small radio antennas, each 6.1m in diameter, is being built as a partnership between the SETI Institute and the University of California Radio Astronomy Laboratory. Last October, Paul G. Allen (who provided the funds for the technology development and the first phase of array construction) pushed a silver button and all 42 antennas of the current ATA-42 slewed to point in the direction of the distant galaxy M81. Specialized electronic backend detectors attached to the ATA began making a radio map of that galaxy and simultaneously began SETI observations of HIP48573, a G5V star near M81 on the sky and a distance of 264 light years from Earth. The Allen telescope array will greatly improve the speed of conducting SETI searches over the next few decades, and it will allow a suite of different search strategies to be undertaken. This paper summarizes some of the earliest SETI observations from the array, and describes the search strategies currently being planned.