First Multi-wavelength Campaign on the Gamma-ray-loud Active Galaxy IC 310

MAGIC is a system of two Imaging Air Cherenkov telescopes, both 17 m in diameter, located on the Canary Island of La Palma, Spain. It covers the electromagnetic spectrum in the VHE range from $50$ GeV to $50$ TeV and achieves an angular resolution of $\sim 0.1^{\circ}$ (Aleksić et al. 2016b ).

The observations during the MWL campaign were conducted after the upgrade of the two telescopes in 2011-2012 was completed AleksicHardwareUpgrade (Aleksić et al. 2016a). On the first night of observation, 2012 November 12-13 (MJD 56243.95-56244.11), during 3.7 h of observation, MAGIC detected a bright flare as reported in Aleksić et al. (2014c ). Further observations until 2013 January 17 (MJD 56309.1) have been carried out as part of the MWL campaign during dark and moon time. Data affected by non-optimal weather conditions were discarded. Only observations during dark night and with moderate moon light were selected, and the standard analysis (Aleksić et al. 2016b ) can be applied. After the selection, the data set consists of $\sim 39$ h including the data of the flaring night. The data cover the zenith distance range of $11^{\circ}<\mathrm{Zd}<\,56^{\circ}$.

The analysis of the data is performed analogously to Aleksić et al. (2014c ) and Aleksić et al. (2016b ). The image cleaning is performed using the dynamical sum-cleaning algorithm presented in Sitarek et al. (2013). The significance of the signal is calculated from Eq. 17 of Li & Ma (1983) using four background regions that do not overlap with the emission from NGC 1275, which is a VHE object $0.6^{\circ}$ away from IC 310. Flux and differential upper limits are calculated according to Rolke et al. (2005) using a 95% confidence level. Following Aleksić et al. (2016b ), we consider for the spectra the following systematic errors: 11% for the flux normalization, 15% for the energy scale and 0.15 for the photon index. As reported previously in Aleksić et al. (2014a , 2014c ), the absorption due to the extragalactic background light (EBL) is only marginal for IC 310. The intrinsic spectra presented in this paper are calculated using the model of Domínguez et al. (2011).

Fermi was launched in 2008 June and since 2008 August 5, it is operated primarily in sky survey mode, scanning the entire sky every three hours atwood (Atwood et al. 2009). The Fermi-LAT is a pair-conversion telescope sensitive to photons between 20 MeV and several hundred GeV ackermann12 (Ackermann et al. 2012).

We calculated spectra using the Fermi Science Tools (v10r0p5)²²2Available online at http://fermi.gsfc.nasa.gov/ssc/data/analysis/software/. for the time period 2012 November 01 (MJD 56232) to 2013 January 31 (MJD 56323). We used the Pass 8 data, the recommended P8R2_SOURCE_V6 instrumental response functions, the isotropic diffuse background template iso_P8R2_SOURCE_V6_v06 and the Galactic emission model gll_iem_v06 acero2016 (Acero et al. 2016)³³3http://fermi.gsfc.nasa.gov/ssc/data/access/lat/BackgroundModels.html, a region of interest (ROI) with a radius of 10^∘ and an energy range of 1 GeV - 300 GeV. We ran the unbinned likelihood analysis with a 90^∘ zenith angle cut to reduce contamination from the Earth limb. As an input for the likelihood analysis we used the 3FGL model acero2015 (Acero et al. 2015). For sources within the ROI the photon index and prefactor (and equivalent for other models) were left free. Spectral upper limits are calculated with a limiting Test Statistic (TS) of 25 mattox1996 (Mattox et al. 1996), and taking into account sources in the spectral model of a radius of 20^∘. The predicted number of counts is low ($\sim$10). Because NGC 1275 appears to be very bright and close to IC 310 with an offset of $0.6^{\circ}$, the results reported here are calculated for energies higher than 1 GeV to mitigate the effects of a larger point-spread function at lower energies.⁴⁴4The plot of the point-spread function can be found online at https://www.slac.stanford.edu/exp/glast/groups/canda/lat_Performance.htm.

The International Gamma-Ray Astrophysics Laboratory (INTEGRAL) satellite is in operation since late 2002 winkler03 (Winkler et al. 2003). It is equipped with several instruments in the hard X-ray to soft gamma-ray range: the high resolution spectrometer SPI at 20 keV–8 MeV vedrenne03 (Vedrenne et al. 2003), and two high angular resolution gamma-ray imagers, called IBIS, which operate at 15–1000 keV and 0.175–10.0 MeV Ubertini03 (Ubertini et al. 2003).

The IBIS data are extracted using the Offline Science Analysis tool OSA, version 10.1. All data between 2012 August and 2013 February are taken into account, where IC 310 was located within $14^{\circ}$ from the pointing center, resulting in 451 science windows. These data are filtered for the energy range between 20 keV and 200 keV. No significant signal is detected from these data and upper limits are derived. SPI data are not used for this publication.

The Swift satellite was launched in late 2004 gehrels04 (Gehrels et al. 2004). Swift provides measurements with telescopes covering the optical and X-ray (soft and hard) ranges. Continuous observations in the hard X-ray range (15–150 keV), mainly for detecting gamma-ray bursts, are provided by the Burst Alert Telescope (BAT). The X-Ray Telescope (XRT) operates in the soft X-ray regime from 0.2 to 10 keV burrows (Burrows et al. 2005).

We extract a spectrum in the energy band of 20–100 keV from the 104-month Swift-BAT survey maps and fit the spectrum with a simple power law with fixed normalization. A fitting statistic for Poisson-distributed source count rates and Gaussian-distributed background count rates as recommended in the XSPEC statistics appendix is used.⁵⁵5https://heasarc.gsfc.nasa.gov/xanadu/xspec/manual/XSappendixStatistics.html The calculations of the flux, photon index, and the corresponding 90% uncertainties are implemented using Monte Carlo simulations of the spectrum.

Furthermore, observations of IC 310 with the Swift-XRT with a total exposure of 45.8 ks were performed in 2012 November and December. The results presented here are compared to earlier observations of 13.2 ks taken in January of the same year. The XRT data were reduced with standard methods, using the most recent software packages (HEASOFT 6.15.1 3) and calibration databases. Spectra were grouped to a minimum signal-to-noise ratio of 5 to ensure the validity of $\chi^{2}$ statistics. For the broad-band SED we applied another re-binning in order to increase the significance of individual points. Spectral fitting was performed with ISIS 1.6.2 houck00 (Houck & Denicola 2000). We fitted the 0.5–10 keV energy band with an absorbed power-law model, which yielded a good fit probability. X-ray data were de-absorbed using abundances from Wilms et al. (2000) and cross sections from Verner et al. (1996). As previous Chandra observations (Aleksić et al. 2014a ) revealed a $N_{\mathrm{H}}$ significantly above the Galactic $N_{\mathrm{H}}$ value of $0.12\times 10^{22}$cm^-2 for IC 310 (Kalberla et al. kalberla10 (2010)) we left $N_{\mathrm{H}}$ free in the fit.

In addition to the X-ray instruments, Swift is equipped with the UltraViolet/Optical Telescope (UVOT) providing observations in the ultraviolet (UV) and optical ranges simultaneous with the XRT gehrels04 (Gehrels et al. 2004). The telescope is equipped with the following filters: V (547 nm), B (439 nm), U (347 nm), UVW1 (260 nm), UVM2 (225 nm), and UVW2 (193 nm). Swift-UVOT data were extracted following standard methods.⁶⁶6http://swift.gsfc.nasa.gov/analysis/UVOT_swguide_v2_2.pdf

The Kungliga Vetenskaps Akademien (KVA) telescopes are located at the Observatorio del Roque de los Muchachos on the island of La Palma, Spain, and are operated by the Tuorla observatory.⁷⁷7http://users.utu.fi/kani/1m They consist of two optical telescopes with mirror diameters of 60 cm and 35 cm. Filters in the R-band (640 nm), B-band (550 nm), and V-band (440 nm) are available. The photometric observations were conducted with the R-band filter and the 35 cm telescope during the MAGIC observations. The data were analyzed using a standard semi-automatic pipeline. The brightness of the source is measured using differential photometry with a standard aperture radius of 5.0^′′. The host galaxy emission is expected to be constant. Any AGN variability would affect the light curve.

Optical, infrared, and ultraviolet data were de-reddened using the same absorbing columns obtained from the Swift-XRT data (Nowak et al. 2012, and references therein).

The 40 m Owens Valley Radio Observatory (OVRO, California, USA) telescope provides radio data for a list of AGNs at 15 GHz nearly twice per week since 2008.⁸⁸8http://www.astro.caltech.edu/ovroblazars/ Details on the observing strategy and the calibration procedures are summarized in Richards et al. (2011). The data presented in this paper cover the time range from 2012 October 31 (MJD 56231) to 2012 December 22 (MJD 56283).

The European VLBI Network is a consortium of several radio-astronomical institutes and telescopes from Europe, Asia, and South Africa.⁹⁹9http://www.evlbi.org/ Due to the large collection area of its telescopes, the EVN provides excellent sensitivity to weak emission. For IC 310, observations in October and November 2012 at the frequencies 1.7, 5.0, 8.4, and 22.2 GHz were carried out.

The MOJAVE project is a long-term VLBI monitoring program at 15 GHz conducted with the Very Long Baseline Array (VLBA) as a continuation of the VLBA 2 cm survey, e.g., Lister et al. (2016).¹⁰¹⁰10https://science.nrao.edu/facilities/vlba The array consists of ten identical 25 m (in diameter) antennas with a baseline up to 8000 km. IC 310 was included in the target list of MOJAVE in early 2012.

A description of the analysis of the EVN data can be partially found in Aleksić et al. (2014c ). More information on the data analysis procedure of the MOJAVE and OVRO data and an extended analysis of these data will be published in a separate paper (Schulz et al. in prep.), which will include additional observations made over a longer period of time that will allow the investigation of possible changes of the jet structure. Only the results of the OVRO observations during the campaign and EVN and MOJAVE flux density measurements will be presented here because they are relevant for the study of the multi-band light curve and the SED.

We also consider historical data of IC 310 for the SED from the Wide-Field Infrared Survey Explorer (WISE) and the Two Micron All Sky Survey (2MASS). The AllWISE Source Catalog wright10 (Wright et al. 2010) covers a time range from 2010 January to 2010 November, while the 2MASS catalog skrutskie06 (Skrutskie et al. 2006) covers from 1997 June to 2001 February. The data are de-reddened using the absorption column derived from spectral fits to the X-ray observations.

During 2012 November to 2013 January (MJD 56245.0–56309.1), IC 310 was detected by MAGIC using data (excluding the flare data of MJD 56244.0) over an effective time (observation time minus dead time) $t_{\mathrm{eff}}=35.3$ h with a significance of 5.22 $\sigma$ above 300 GeV. This low significance measured over rather a long time span already indicates a low flux state after the flare. Here we have excluded the flare data from the data selection to test the detection of the object outside of the flaring state and because a large part of the observations used for the MWL SED, e.g., in X-rays, was performed not simultaneous to the TeV flare.

The energy spectrum $\left(\mathrm{d}N/\mathrm{d}E\right)$ between 82 GeV and 2.1 TeV during the campaign from 2012 November to 2013 January (Fig. 1), excluding the flare data, can be described with the power-law function:

with a flux normalization $f_{0}=(3.12\pm 0.91_{\mathrm{stat}}\pm 0.34_{\mathrm{syst}})\times 10^{-13}\,\mathrm{TeV}^{-1}\,\mathrm{cm}^{-2}\,\mathrm{s}^{-1}$  at 1 TeV and a photon spectral index of $\Gamma=(2.36\pm 0.30_{\mathrm{stat}}\pm 0.15_{\mathrm{syst}})$. The normalization energy of 1 TeV was fixed and selected for easier comparison with previous measurements. As simultaneous Swift-XRT and UVOT data are only available for the 2012 November and December observations, we also calculated a MAGIC spectrum for the time period 2012 November to December in the same energy range. The resulting spectrum agrees, within the errors, with the spectrum calculated for the entire campaign. Thus, in the subsequent sections we only show and discuss the spectrum for the entire campaign.

We further investigate the change in the spectrum at different flux states by comparing our results with previous measurements, as well as by studying spectra of different flux states during the flare on 2012 November 12-13. In Aleksić et al. (2014a ) and Aleksić et al. (2014c ), no significant spectral variability was reported. Here, we present individual spectra during the flare. The division of the data of the flare according to different flux states is given in Table 1 and shown in Fig. 2. Besides the separations based on the fluxes, the definitions of the intervals are also determined by the duration of observing runs of typically 20 minutes. The resulting spectra are shown in Fig. 3 and their parameters are listed in Table 2.

In Fig. 4, we present for all temporally non-overlapping data the photon spectral index versus the integrated flux (overlapping data would include a bias so the averaged flare spectrum on 2012 November 12 - 13 is not included). All integrated fluxes represent mean fluxes and have been obtained by fitting all light curves above 300 GeV with a constant line, both the light curves from previous measurements (Aleksić et al. 2014a ) and from the data presented here. A constant fit to the data points in Fig. 4 reveals a $\chi^{2}/\mathrm{d.o.f}$ of 52.5/7, thus a low probability for being constant of $4.6\times 10^{-9}$. A higher probability can be obtained with a linear function which yields a $\chi^{2}/\mathrm{d.o.f}$ of 7.0/6 and a probability of $0.32$.

The mean integrated flux over the entire period is measured to be $F_{\mathrm{mean}}=(1.59\pm 0.29)\times 10^{-12}$ cm^-2 s^-1 above 300 GeV when excluding the data from the 2012 November flare. This is $\sim 40$ times lower than the mean integrated flux of $(6.08\pm 0.29)\,\times 10^{-11}$ cm^-2 s^-1 reported for the 2012 November flare in Aleksić et al. (2014c ). Due to the faint emission, a monthly light curve is computed. The light curve is calculated assuming a simple power-law distribution with a photon index of $\Gamma=2.4$. For the data during the flare, a different photon index, $\Gamma=2.0$, is used. The monthly calculated light curve is shown in Fig. 9. Fitting this light curve with a constant reveals a flux of $F_{\mathrm{const.}}=(1.42\pm 0.29)\times 10^{-12}$ cm^-2s^-1 with a $\chi^{2}$/d.o.f. of 3.2/2 (probability of 0.20). Thus, no significant variability of the flux is found from month to month. Results from individual days are given in Table 3.

In the time range from 2012 November 01 (MJD 56232) to 2013 January 31 (MJD 56323), IC 310 could not be detected with Fermi-LAT (TS$<25$) over the entire three months period. The measured light curve is shown in Fig. 9. Only upper limits of the flux and one flux point in 2013 January could be calculated. Thus, no further conclusion can be drawn on the variability behaviour of IC 310. We also searched for individual gamma-ray event candidates in the Fermi-LAT detector during the MWL campaign. Within a circle with a radius of $10.0^{\circ}$ around IC 310, only four events could be found above 1 GeV with a larger likelihood to originate from IC 310 than from NGC 1275 and a probability of higher than 50%. The probability for each photon was calculated using gtsrcprob. They were detected on MJD 56248.4, 56298.8, 56303.7, and 56318.1, with energies of 4.8, 8.0, 2.4, and 17.2 GeV, respectively. The arrival times are indicated with blue lines in Fig. 9. The calculated 95% confidence level upper limits for the SED are shown in Fig. 10. The blue upper limit spectrum covers the full time range of the campaign from 2012 November 01 to 2013 January 31 and an energy range from 1 GeV to 72 GeV.

The light curve measured with Swift-XRT is shown in Fig. 5 and in Fig. 9. An overview of the results - e.g., the flux level in the energy range of 2–10 keV, the photon index, and $N_{\mathrm{H}}$ - can be found in Table 4.

The temporal evolution of the flux in the energy range 2–10 keV, the photon index, and the column density $N_{\mathrm{H}}$ in 2012 November to December during the campaign is presented in Fig. 5. The mean energy flux has been measured to be $(0.61\pm 0.01)\times 10^{-11}$ erg cm^-2 s^-1, which is about five times higher than during previous measurements (Aleksić et al. 2014a ) and moderately higher (factor of 1.4) than in 2012 January (see Table 4). A fit to the light curve with a constant line reveals a probability of $7.3\times 10^{-7}$ for a constant energy flux of $(0.58\pm 0.01)\times 10^{-11}$ erg cm^-2 s^-1 ($\chi^{2}$/d.o.f.$=49.6/11$). The observation on MJD 56280.82 deviates by $\sim 5\,\sigma$ from this constant energy flux value.

Comparing the light curve with the temporal evolution of the photon index yields evidence for a spectral hardening with increasing energy flux. We fit the photon index versus time with a constant, yielding $\chi^{2}$/d.o.f.$=45.2/11$ (probability of $4.5\times 10^{-6}$) indicating a change of the photon index from day to day. This evidence is also found when displaying the photon index as a function of the energy flux between 2–10 keV as shown in Fig. 6. A linear fit gives a $\chi^{2}$/d.o.f. of $14.0/10$ (probability of 0.17). Thus, a harder-when-brighter behaviour is observed during the campaign. Although spectral and flux variability in the X-ray band were previously reported in Aleksić et al. (2014a ), a harder-when-brighter trend for these observations was not found.

The hydrogen column density stayed constant during the campaign ($\chi^{2}$/d.o.f.$=13.4/11$, probability of 0.26 for a constant fit) and is consistent with the Galactic value for IC 310 ($0.12\times 10^{22}$ cm^-2) from Kalberla et al. (kalberla10 (2010)) taking into account the large systematic uncertainties ($\sim 30\%$) of the survey. Hence, no conclusions can be drawn on intrinsic photo-absorption during the campaign.

For the investigation of the multi-wavelength SED, all data taken in 2012 November and December were combined to derive an averaged spectrum during the campaign. The result is shown in Fig. 7. It can be well described with a simple absorbed power-law in the range 0.5–10 keV with a hard photon spectral index of $\Gamma=2.036^{+0.022}_{-0.019}$. This index is comparable within the errors with the index of the averaged spectrum obtained from measurements at the beginning of the year 2012.

From the INTEGRAL data, a 1 $\sigma$ upper limit is extracted from the variance of the stacked mosaic image at the source position taking into account a photon index of 2.0, which is extrapolated from the Swift-XRT band. This results in an upper limit of the energy flux of $2.9\times 10^{-12}$ erg cm^-2 s^-1 at 110 keV in the energy range 20–200 keV. Assuming a softer photon index of 2.5, the upper limit yields $1.7\times 10^{-12}$ erg cm^-2 s^-1 at 110 keV in the energy range 20–200 keV. For the broad band SED in Fig. 10, we show the 2 $\sigma$ upper limit calculated for the photon index of 2.0 and multiply it with a factor of 1.5 for the root mean square of the significance map.

The BAT data yield an energy flux of $(3.2\pm 1.4)\times 10^{-12}$ erg cm^-2 s^-1 for the energy range 20–100 keV and a photon index of $1.8\pm 1.1$ for the 104-month Swift-BAT survey maps with a signal-to-noise ratio of 1.85 $\sigma$. Uncertainties are given at a 90% confidence level. For the broad-band SED, we use the energy flux value instead of an upper limit. We apply the criterion of $3\,\sigma$ for the calculation of energy flux upper limits. But, as three times the $1\,\sigma$ energy flux uncertainty is smaller than the energy flux measurement, we do not classify the energy flux as an upper limit.

The KVA R-band optical light curve is shown in Fig. 9 and does not show any signs of variability. The light curve in the time range of 2012 November 12 to 2013 February 02 (MJD 56243–56325) is consistent with a constant flux density of $(9.08\pm 0.04)$ mJy ($\chi^{2}$/d.o.f.$=5.3/17$, probability of $0.99$). As no further historical monitoring in the R-band was conducted for IC 310, the flux density cannot be compared with other measurements.

An overview of the observations from 2012 November to December by Swift-UVOT is given in Table 5 together with the 2012 January measurements. The results show no significant variability for any of the filters (Fig. 8). A constant fit to the 2012 November to December data revealed no variability for B, U, V, UVW1 with a fit probability of 0.98, 0.99, 0.95, and 0.98, respectively. Smaller fit probabilities were obtained for the UVM2 and UVW2 measurements (0.26 and 0.03). Comparing the measurements in 2012 January and 2012 November-December, only at the highest frequencies (UVM2 and UVW2) the mean flux densities from late 2012 are not compatible within the errors with the measurements in early 2012.

To interpret and model the broad band SED, mean flux densities for each frequency were calculated from the data taken in 2012 November to December. The values obtained are listed in Table 6.

The SED in Fig. 10 shows the observed and de-reddened KVA and mean data from Swift-UVOT using the absorption measurement from Swift-XRT.

The light curve at 15 GHz obtained by the OVRO monitoring program is shown in Fig. 9. From a fit of the data points from 2012 October 30 to December 23 (MJD 56230-56284) to a constant, a mean flux density of $(0.151\pm 0.002)$ mJy is measured with a $\chi^{2}$/d.o.f. of $31.1/13$ and rather low probability of $0.003$ for being constant.

From the EVN and MOJAVE data, the core and total flux density were calculated and included in Fig. 10. The estimated uncertainties for the EVN flux density measurements at 1.7, 5.0, 8.4 GHz are 10% and for 22 GHz, 15%. For the MOJAVE data we adopt an uncertainty of 5% for the total flux density lister2005 (Lister & Homan 2005). We use the results from the 5.0 GHz observation taken from Aleksić et al. (2014c ).

The combined multi-wavelength light curve is shown in Fig. 9. An exceptional TeV flare was found by MAGIC in November 2012. IC 310 remained active at VHE even after the flare. Only Fermi-LAT, Swift-BAT and partially KVA observed simultaneous to MAGIC on 2012 November 12-13. However, IC 310 could not be detected with the first two instruments during the TeV flare and the optical emission observed by KVA is dominated by the host galaxy.

Contemporaneous measurements starting after the flare indicated a high and variable state of the object in the soft X-ray range. This marks the first time that a contemporaneous measurement of the VHE and the soft X-ray flux is reported for IC 310.

Conclusions on a correlation between the two bands should be drawn with caution because historical simultaneous measurements in both energy ranges are missing. However, considering the harder-when-brighter trend in the VHE and X-ray bands reported in Fig. 4 and Fig. 6, one can speculate that these two bands may be connected.

A harder-when-brighter behaviour is frequently observed for other sources such as high-frequency peaked BL Lac (HBL) objects (Pian et al. (1998); Giommi et al. (2000); Aleksić et al. (2013); Furniss et al. (2015); Aleksić et al. (2015); Kapanadze et al. (2016); Ahnen et al. 2016a ; Baloković et al. (2016)). Here, the increase of the X-ray and VHE flux is combined with a hardening of the spectrum. This is consistent with the synchrotron self-Compton (SSC) mechanism, see Sect. 4.2.2. The first hump in the SED moves towards higher frequencies. The same happens for the second hump, as during a higher flux state the synchrotron photons in the low energy band are seeds for the inverse-Compton emission observed in the high-energy bands. If such a behaviour is observed for a blazar with a small viewing angle, one should expect the same also for radio galaxies. This would be in-line with the unified model by Urry & Padovani (urry95 (1995)). Note that X-ray data from other misaligned blazars (e.g., NGC 1275) often show complex emission, with a non-pure power-law and extended jet emission or X-ray emission of the filaments fabian2011 (Fabian et al. 2011).

The optical light curve does not show a significantly higher flux or variability. The lack of optical variability is consistent with the result of the SED modeling (see Sect. 4.2.3). The optical-infrared emission is ascribed entirely to the host galaxy, while the variable jet emission starts to dominate in the far-UV and soft X-ray band. The radio light curve indicates weak variability. As radio flares combined with an appearance of a new radio component in the VLBI images are sometimes found a few months after a gamma-ray flare Acciari et al. (2009), the time period covered here is too short to draw conclusions.

Generally, the VHE flare might be interpreted as an injection of fresh electrons and positrons into the SSC emission region. One possible origin of these particles could be the electro-magnetic cascades in the gap region of magnetospheric models resulting in an increased number of $e^{+}e^{-}$-pairs (Beskin et al. (1992); Rieger & Mannheim (2000); Neronov & Aharonian (2007); Levinson & Rieger levinson (2011)). Such a model has been used to explain the flaring activity from M 87 by Levinson & Rieger (levinson (2011)) as well as the minute time-scale variability of the gamma-ray emission of IC 310 at the beginning of the MWL campaign reported by Aleksić et al. (2014c ) and critically examined in Hirotani & Pu (2016). Since the magnetic field is assumed to be along the jet axis, these particles should then move along the jet and maybe also into the emission region which we observed after the flare. This could cause a higher flux in the X-ray regime as well as at VHE. Note that the quiescent state of IC 310 might be undetectable with MAGIC as sometimes IC 310 could not be detected over a long observation time (Aleksić et al. (2012)). Furthermore, these fresh particles moving along the jet could also explain a general activity in the radio band.