ArticlePDF Available

Janus: A Combined Radar and Vibration Sensor for Beehive Monitoring

Authors:
  • MaineBiosensors LLC

Abstract and Figures

A novel beehive monitoring sensor with two faces is described. This sensor is attached to the outside of a hive, near the hive entrance. The outward-looking sensor is a 24 GHz continuous-wave Doppler radar for monitoring bee flying activity. The inward-looking sensor is a piezo-electric transducer. Unlike a conventional microphone that would pick up the sounds bees make, the piezo-electric transducer picks up the incidental vibrations transmitted by bee activity to the hive structure. The root-mean-square powers in concurrent radar and vibration measurements are shown to be highly correlated during honeybee swarming and robbing events. Principal component analysis was applied to radar, vibration, and environmental measurements to further reduce false alarms.
Content may be subject to copyright.
VOL. 5, NO. 3, MARCH 2021 1500204
Sensor phenomena
Janus: A Combined Radar and Vibration Sensor for Beehive Monitoring
Herbert M. Aumann1* , Margery K. Aumann2, and Nuri W. Emanetoglu1**
1Department of Electrical and Computer Engineering, The University of Maine, Orono, ME 04469 USA
2Dover-Foxcroft, ME 04426 USA
*Life Member, IEEE
**Member, IEEE
Manuscript received January 8, 2021; revised January 28, 2021; accepted January 31, 2021. Date of publication February 3, 2021; date of current version
February 26, 2021.
Abstract—A novel beehive monitoring sensor with two faces is described. This sensor is attached to the outside of a hive,
near the hive entrance. The outward-looking sensor is a 24-GHz continuous-wave Doppler radar for monitoring bee flying
activity. The inward-looking sensor is a piezoelectric transducer. Unlike a conventional microphone that would pick up
the sounds bees make, the piezoelectric transducer picks up the incidental vibrations transmitted by bee activity to the
hive structure. The root-mean-square powers in concurrent radar and vibration measurements are shown to be highly
correlated during honeybee swarming and robbing events. Principal component analysis was applied to radar, vibration,
and environmental measurements to further reduce false alarms.
Index Terms—Sensor phenomena, beehive monitoring, Doppler radar, piezoelectric vibration transducer.
I. INTRODUCTION
There are two unusual events in an apiary that require immediate
intervention by the beekeeper: swarming and robbing.
Swarming is a natural event during which about half the bee popu-
lation leaves the hive to establish a new colony. In as much as possible,
beekeepers like to avoid swarming, and its associated loss of honey
production, using proper management techniques [1]. When swarming
happens unexpectedly, however, the beekeeper would like to be alerted
as soon as possible. Swarming may take place in a matter of minutes.
It usually starts with a great uproar from the beehive, followed by
a rapid disgorgement and flight of thousands of bees in a cloud. The
swarm initially lands and forms a tight cluster close to the original hive
before flying off to its final destination. This bivouac may last for as
little as an hour, or depending on the weather, it may last for days [2].
Quick action during the bivouac phase is the beekeeper’s best chance
of capturing a swarm and thereby increasing his apiary stock [3].
At times of poor honey flow, bees from a stronger hive may attack a
weaker hive and steal its honey [4]. If successful, robbers can clean all
the honey out of a hive in a few days. Robbing tends to be a persistent,
all-day event that is usually accompanied by a high level of noise and
continuous frenzied flying activity in front of the hive [3]. One way of
dealing with that situation is either to reduce temporarily the size of
the hive entrance or to close it altogether [3].
A commonly occurring and harmless activity is the “mass orienta-
tion flight,” a bee activity that has so far eluded a concise biological
explanation [5]. In this scenario, a substantial cloud of bees flies back
and forth in front of the hive entrance. Mass orientation flights have
often been mistaken for swarming, except that the bees do not fly far
away from the hive, and they return to the hive after a few minutes.
The level of flying activity of orientation flights and swarming can
be very similar; however, the level of acoustic noise associated with
Corresponding author: Herbert M. Aumann (e-mail: herbert.aumann@maine.
edu).
Associate Editor: S.-R. Kothapalli.
This work was supported by the Entirely Self-Funded by the Authors.
Digital Object Identifier 10.1109/LSENS.2021.3056870
Fig. 1. Radar/vibration sensor installed on beehive 8B.
orientation flights is considerably lower. Although orientation flights
are also detected by our sensor, they usually do not require intervention
by the beekeeper.
By carefully noting flight activity near the beehive entrance and
acoustic noise from the hive, an experienced beekeeper can tell very
quickly if the bees are happy, sad, or mad [6]. However, it is often not
practical for the beekeeper to check on hives in a remote apiary every
day. An inexpensive beehive monitoring system is of interest to that
would warn the beekeeper that a swarm is about to take place, or has
just occurred, or that a robbery is in progress.
An external sensor, as shown in Fig. 1, can meet the aforementioned
requirement. We will show that the observations of swarming events,
done both with a Doppler radar and with a vibration sensor, are
highly correlated, even though they are based on entirely different
phenomenology. When a major hive perturbation occurs, and the
Doppler and vibration levels are both high, then it is likely to be a
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
1500204 VOL. 5, NO. 3, MARCH 2021
Fig. 2. Radar/vibration sensor block diagram.
swarm. When the levels differ significantly, then the event is most
likely due either to robbing or to an “orientation flight.”
We review prior related work in Section II. The sensor hardware
is described in Section III. In Section IV, we verify the operation of
the two-faced sensor, and in Section V, we show some experimental
results.
II. PRI OR WORK
The unique sounds that bees make have been described since an-
tiquity [3]. We are all familiar with the buzz of bee wings, which is
typically at 270 Hz. At other times, one can hear intermittent sounds,
such as queen bees tooting (550 Hz) or quacking (300 Hz), or workers
piping (400 Hz) [7]–[9]. Most bee sounds are well below 600 Hz.
Numerous sensors have been developed for collecting, remotely
and in real-time, acoustic signals from bees within a beehive [10]–
[17]. A review of such sensors can be found in [17]. The installation
requires opening the hive and inserting a microphone between the
frames. Eventually, the microphone becomes covered with propolis
and thereby rendered less effective. Interpretation of the bee sounds
is left to the beekeeper or else requires expensive offline equipment
for the needed feature extraction. Sounds captured with an acoustic
microphone outside the hive [17] while easier to achieve have often
been found to be corrupted by environmental interference.
Bee behavior has also been studied by measuring localized vibra-
tions with an accelerometer. In such studies, the accelerometer was
installed either inside the hive [4], [18] or in the hive wall [19].
In an entirely different approach, we demonstrated the use of a low-
power continuous-wave (CW) Doppler radar to listen to bee sounds
[20] or to monitor flying activity outside the beehive [21]–[23]. In
this letter, we combine this radar sensor with a vibration sensor in a
two-faced apparatus to improve swarm detection.
III. JAN US SE NSOR
We have named our sensor “Janus” after the god of Roman mythol-
ogy. Janus is usually pictured with two, sometimes dissimilar faces,
one looking forward and one looking backward. Similarly, our system
contains two sensors, looking in opposite directions. A block diagram
of the system is shown in Fig. 2. One sensor monitors bee activity
inside the beehive (red), and one sensor monitors activity outside the
beehive (blue). A Janus sensor implementation is depicted in Fig. 3.
A. Vibration Sensor
The vibration sensor is based on a very inexpensive 26-m-diameter
piezoelectric transducer commonly used in acoustic guitar pickups. It
is firmly attached to the outside wall of a wooden Langstroth hive above
Fig. 3. Janus front end.
Fig. 4. Vibration sensor impulse response.
Fig. 5. Relative signal strength of vibration (red) and radar (blue)
sensors after amplification.
the hive entrance. The transducer is connected to an audio amplifier
with about 70-dB gain and a 300–2000 Hz frequency response, as
showninFig.5inred.
Careful observations reveal that the frequency of the signal mea-
sured by such a sensor installation is primarily determined by the
resonant characteristics of the wooden hive and its content. It has very
little to do with the classic “bee sounds” described in Section II. It
is a by-product of thousands of bee feet “strumming the guitar, as it
were. We verified the response of an empty hive by striking it with a
suspended ball bearing. The data processing is described in Section IV.
Consistent with the theoretical acoustic cutoff frequency, a completely
empty hive has a resonance at about 360 Hz. We then repeated the test
with a hive filled with ten empty frames. As illustrated in Fig. 4, the
resonant frequency of the full hive has shifted to about 1100 Hz.
VOL. 5, NO. 3, MARCH 2021 1500204
Fig. 6. Vibration and radar FTI plots.
B. Doppler Radar
The motion detector, shown in Fig. 3, is a 24-GHz CW Doppler
radar found in automobile collision avoidance systems. This sensor
has been used successfully to observe bee flying activity [23]. No
adverse radiation effects on bees were observed from an active cell
phone inserted into a bee hive [24]. The present radar is outside the
hive and transmits 1/100 power of a cell phone. A similar 10.5-GHz
sensor was described in [22]. However, due to the much higher radar
cross section of bees at 24 GHz, such a radar has significantly better
sensitivity [22].
The departure of foraging bees from the hive entrance causes the
Doppler frequency of the 24-GHz CW radar to shift from 0 Hz to about
800 Hz over a distance of about 2 m [23]. When using the same audio
amplifier as above, the corresponding 1/R4losses distort the effective
signal amplification, as shown in Fig. 5 in blue.
IV. DATA COLLECTION AND PROCESSING
Two different data collection and processing systems were used with
the same Janus front end.
A. Hard-Wired Sensor
In the initial system, the Janus front end was directly connected from
a beehive to the stereo microphone input of a laptop computer. Raw
12-b Analog-to-digital converter (ADC) samples at 8 ksamples/s were
collected from both, the radar and vibration sensor. These samples
were recorded for 20 s every 2 min from sunrise to sunset. Frequency-
time-intensity (FTI) plots were generated with MATLAB for each 20-s
data collection. An example is given in Fig. 6.
While the radar FTI shows the Doppler of individual bees coming
and going, the vibration FTI from thousands of bees is considerably
more uniform. In Fig. 7, we averaged the FTI’s of both channels over
the 20-s data collection. The plot shows remarkable similarity to the
expected frequency response shown in Fig. 5.
It was shown in [21] that the root-mean-squared (rms) value of the
raw Doppler signal is equivalent to the total power in the Doppler
spectrum and is a good indicator of the level of bee activity. We
similarly calculated the rms value of the vibrational signal. The rms
Fig. 7. Average frequency response of Fig. 6.
Fig. 8. Remote bee activity sensor.
values in Fig. 7 were calculated directly from raw ADC samples
without spectral analysis.
B. Remote Sensor
The second sensor version, for remote data collection from multiple
hives, was built with commercial break-out boards, as shown in Fig. 8.
Here, a solid-state timer actuated a microprocessor approximately
every 2 min. The microprocessor then collected 2 s of 12-b radar and
vibration samples and calculated their rms values. These data were
time tagged and transmitted over a 433-MHz wireless link to a base
station for recording on a laptop computer. Up to nine sensors could
be monitored from a distance of almost 200 m. Since the transmission
duty factor is only 0.02%, message collisions occurred only very rarely
and were easily identified by checksum error.
V. EXPERIMENTAL RESULTS
Radar and vibration measurements with identical wireless sensors
were made on three beehives, identified as 6B, 7B, and 8B, located in
Dover-Foxcroft, ME, USA. The hives were continuously monitored
from July 2020 to October 2020.
A. Daily Measurements
Representative examples of days with unusual events are shown in
Fig. 9. All events were visually confirmed. Radar and vibration rms
levels for a swarming event [see Fig. 9(a)] were very high and highly
correlated (ρ>0.5). No significant lag between vibration and radar
data could be observed. As expected, the vibrational level was higher
during a robbing situation [see Fig. 9(b)] and lasts for most of the day.
Although the radar signals in Fig. 9(a) and (c) are similar, correlation
between radar and vibrational signals for “orientation flights” was very
low.
1500204 VOL. 5, NO. 3, MARCH 2021
Fig. 9. Examples of (a) swarming, (b) robbing, and (c) orientation
flights.
Fig. 10. PCA analysis of hive radar and vibration data.
B. Seasonal Measurements
Observations over the course of the season indicated that unusual
activities did not always match the ideal cases illustrated in Fig. 9,
leading to false alarms. Other factors, such as ambient temperature and
solar radiation [22], also influence the level of activity. We explored
the correlation between radar, vibration, temperature, and solar flux by
principal component analysis (PCA). The PCA analysis was carried
out in MATLAB on a laptop attached to the base station. Each vertical
line in Fig. 10 represents the magnitude of the most significant principal
component derived from daily data, as exemplified in Fig. 9.
As an experiment, for one hive (8B), the space available for bee pop-
ulation expansion was limited on purpose to induce swarming. Indeed,
the hive swarmed five times. PCA analysis clearly and unambiguously
identified these events, allowing the swarms to be captured. We noticed
a reduced level of activity about ten days before the primary swarm.
This reduced activity level has been reported elsewhere [25], but it is
difficult to detect automatically or in advance.
VI. CONCLUSION
Both the radar and vibration sensor, when mounted on the outside
wall of a beehive, are capable of detecting swarming and robbing activ-
ity. While the vibration sensor by itself is considerably less expensive,
the detection performance can be improved (i.e., false alarms reduced)
by combining measurements from both sensors and subjecting them
to a PCA.
REFERENCES
[1] V. Vickery, The Honey Bee: A Guide for Beekeepers. Westmount, QC, Canada:
Particle, 1991.
[2] S. Repasky, Swarm Essentials, Ecology, Management, Sustainability. Kalamazoo,
MI, USA: Wicwas, 2014.
[3] R. A. Jones and S. Sweeney-Lynch, The Beekeeper’s Bible.NewYork,NY,USA:
Abrams, 2011.
[4] R. Conrad, Natural Beekeeping. White River Junction, VT, USA: Chelsea, 2007.
[5] J. Tautz, The Buzz About Bees. Berlin, Germany: Springer-Verlag, 2009.
[6] H. Storch, At the Hive Entrance. Brussels, Belgium: Eur. Apicultural Ed., 2019.
[7] H. Eren, L. Whiffler, and R. Manning, “Electronic sensing and identification of
queen bees in honeybee colonies,” in Proc. IEEE Instrum. Meas. Technol. Conf.
Sens., Process., Netw., 1997, vol. 2, pp. 1052–1055.
[8] C. Collison, “A closer look: Piping, tooting, quacking,” Bee Culture, Eastern Api-
cultural Soc., Medina, OH, USA, 2018, pp. 1–8.
[9] A. Terenzi, S. Cecchi, and S. Spinsante, “On the importance of the sound emitted
by honey bee hives, Veterinary Sci., vol. 7, pp. 168–174, 2020.
[10] S. Cecchi, S. Spinsante, A. Terenzi, and S. Orcioni, “A smart sensor-based measure-
ment system for advanced bee hivemonitoring,” Sensors, vol. 20, 2020, Art. no. 2726.
[11] F. E. Murphy, B. Srbinovski, M. Magno, E. M. Popovici, and P. M. Whelan, “An
automatic, wireless audio recording node for analysis of beehives, in Proc. 26th
Irish Signals Syst. Conf., 2015, pp. 1–6.
[12] A. Zgank, “Bee swarm activity acoustic classification for an IoT-based farm service
sensors,” Sensors, vol. 20, 2020, Art. no. 21.
[13] A. Zlatkova, Z. Kokolanski, and D. Tashkovski, “Honeybees swarming detection
approach by sound signal processing,” in Proc. 29th Int. Sci. Conf. Electron., 2020,
pp. 1–3.
[14] S. Gil-Lebrero, F. Quiles-Latorre, M. Ortiz-López, V. Sánchez-Ruiz, V. Gámiz-
López, and J. Luna-Rodríguez, “Honey bee colonies remote monitoring system,”
Sensors, vol. 17, 2017, Art. no. 55.
[15] W. Meikle and N. Holst, Application of continuous monitoring of honeybee
colonies,” Apidologie, vol. 46, pp. 10–22, 2015.
[16] A. T. Nolasco, S. Cecchi, S. Orcioni, H. L. Bear, and E. Benetos, “Audio-based
identification of beehive states,” in Proc. IEEE Int. Conf. Acoust., Speech Signal
Process., 2019, pp. 8256–8260.
[17] V. Kulyukin, S. Mukherjee, and P. Amlathe, “Toward audio beehive Monitoring:
Deep learning vs. standard machine learning in classifying beehive audio samples,
Appl. Sci., vol. 8, 2018, Art no. 1573.
[18] M. Ramsey, M. Bencsik, and M. Newton, “Long-term trends in the honey-
bee ‘whooping signal’ revealed by automated detection,” PloS ONE, vol. 12, 2017,
Art. no. e0171162.
[19] M. Bencsik et al., “The prediction of swarming in honeybee colonies using vibra-
tional spectra,” Sci. Rep., vol. 10, 2020, Art. no. 9798.
[20] H. Aumann and N. Emanetoglu, “The radar microphone: A new way of monitoring
honey bee sounds,” in Proc. IEEE Sensors, 2016, pp. 1–2.
[21] H. Aumann, B. Payal, N. W.Emanetoglu, and F. Drummond, “An index for assessing
the foraging activities of honeybees with a Doppler sensor,” in Proc. IEEE Sens .
Appl . Symp ., 2017, pp. 1–5.
[22] A. Souza Cunha, J. Rose, J. Prior, H. Aumann, N. Emanetoglu, and F. Drummond,
“A novel non-invasive radar to monitor honey bee colony health,” Comput. Electron.
Agriculture, vol. 170, 2020, Art. no. 105241.
[23] H. Aumann, “A technique for measuring the RCS of free-flying honeybees with
a 24 GHz CW Doppler radar,” in Proc. 12th Eur. Conf. Antennas Propag., 2018,
pp. 181–184.
[24] R. Odemer and F. Odemer, “Effects of radiofrequency electromagnetic radiation
(RF-EMF) on honey bee queen developmentand mating success,” Sci. TotalEnviron.,
vol. 661, pp. 553–562, 2019.
[25] S. Ferrari, M. Silva, M. Guarino, and D. Berckmans, “Monitoring of swarming
sounds in bee hives for early detection of the swarming period, Comput. Electron.
Agriculture, vol. 64, pp. 72–77, 2008.
... Other monitoring technologies: Audio, video, and other monitoring technologies such as gas sensors (Szczurek et al., 2019a), thermal imaging (Klein et al., 2014), infrared sensors (Shaw et al., 2010), GPS (Ntawuzumunsi et al., 2021), accelerometers (Kontogiannis, 2019), vibration detectors (Aumann et al., 2021), and light intensity sensors (Keppner and Jarau, 2016) are some less standard technologies we identified for Operational monitoring. Although they provide valuable data, perhaps their cost and complexity inhibits widespread adoption. ...
... Recent approaches to detecting this event include the combination of temperature, sound, hive weight, and humidity Hong et al., 2020;Anand et al., 2018). Researchers have investigated too how vibration (Aumann et al., 2021;Ramsey et al., 2020) and radar sensors (Aumann et al., 2021) can be used to detect swarming events. ...
... Recent approaches to detecting this event include the combination of temperature, sound, hive weight, and humidity Hong et al., 2020;Anand et al., 2018). Researchers have investigated too how vibration (Aumann et al., 2021;Ramsey et al., 2020) and radar sensors (Aumann et al., 2021) can be used to detect swarming events. ...
Article
Full-text available
Hive bees provide essential pollination services to human agriculture. Managed honey bees in particular pollinate many crops, but also create honey and other bee products that are now of global economic importance. Key aspects of honey bee behaviour and colony health can be understood by observing hives. Hence, the limitations of manual observation are increasingly being addressed by new technologies that automate and extend the reach of hive monitoring. Here we propose a framework to classify and clarify the potential for sensor-assisted hive monitoring to inform apiculture and, ultimately, improve hive bee management and bee health. This framework considers hive monitoring approaches across three newly proposed categories: Operational monitoring, Investigative monitoring, and Predictive monitoring. These categories constitute a new “OIP Framework” of hive monitoring. Each category has its own requirements for underlying technology that includes sensors and ICT resources we outline. Each category is associated with particular outcomes and benefits for apiculture and hive health monitoring detailed here. Application of these three classes of sensor-assisted hive monitoring can simplify understanding and improve best-practice management of hive bees. Our survey and classification of hive monitoring to date show that it is seldom practised beyond honey bees, despite the need to understand bumble bees and stingless bees also. Perhaps unsurprisingly, sensor-based hive monitoring is shown to remain primarily a practice of countries with upper-middle and high income economies. Yet we show benefits from which all countries, especially countries with lower-middle and low income economies, stand to gain through improved sensor-based hive monitoring. These include a better understanding of environmental change, an increased ability to manage pollination, an ability to respond rapidly to hive health issues such as pests and pathogens, and even an ability to react quickly to the danger posed to insects and humans alike by extreme events such as floods and fires. Hive health is a critical issue for bees, beekeepers, horticulture and ecosystem sustainability. We show that its assessment requires a rigorous and coherent approach that has not yet been adequately formalised. Finally, we anticipate that the future of hive monitoring lies in the application of Predictive monitoring, such that a hive’s anticipated future state can be preemptively managed by beekeepers working iteratively with novel hive monitoring technologies.
... The benefits of remote prediction of swarm events include inspecting hives only when needed rather than regularly (Aumann et al., 2021). Furthermore, the data collected can indicate when honey flow stops and colonies need to be relocated or fed to maintain health or being able to find sites with higher honey yields . ...
... In Aumann et al. (2021), a similar approach was used to monitor rather simple activity events of flying honey bees. Swarming, robbing and orientation flights could be indicated with root-mean-squared counts, also known as the quadratic mean (Jones, 2018), which correlates with the total power in the Doppler spectrum. ...
Article
Full-text available
For nearly 100 years, electronic bee counters have been developed using various technologies to track the foraging activity of mostly honey bee colonies. These counters should enable remote monitoring of the hives without disturbing natural flight behaviour while generating precise scientific data. However, there are few counters on the market that are able to fulfill this task. One main challenge is the lack of standardized methods to validate a counter's precision, as validation is crucial to categorize and judge the data produced by the counter, especially for scientific purposes. Another challenge is the interpretation of flight data to measure the effects of environmental or anthropogenic sources. Nevertheless, recent developments in the field are promising. This review describes the historic development of automated bee flight measurements and critically compares validation methods to encourage their improvement. To increase the comparability of future analyses of bee counters, current advances in data interpretation are also presented. This article is protected by copyright. All rights reserved.
... The data captured from these kinds of new advanced systems will be processed, and the extracted important features enable beekeepers monitoring beehives remotely without interrupting the bees' natural behavior at their hives. Aumann et al. (2021) introduced a system that combines both radar (for monitoring bee flying activity) and vibration sensor (for picking up incidental vibrations transmitted by bee activity to the hive structure). For many other systems introduced in the literature, one can refer to IoBee, 2 OSBeehives, 3 Hivemind, 4 Gobuzzr, 5 and SAMS (Wakjira et al. 2021). ...
Article
Full-text available
Due to the important role of honey bees in life, there has been increasingly advanced techniques introduced to support apiarists in taking the best care of their bees. Recently, machine learning (ML) methods emerge as a powerful tool among these techniques and have great contributions to automated beehive monitoring systems with low-cost and better performance. By analyzing the data collected from beehives, ML algorithms are able to solve a number of crucial problems in monitoring the beehives such as early detecting the phenomena of swarming, identifying the bee queen's absence, and recognizing pest infestations. In this study, we suggest several techniques to enhance the performance of the machine learning models applied to monitoring the honey beehives. Particularly, we apply an advanced technique for tuning hyper-parameters of the MFCCs and hyper-parameters tuning for bee sound recognition ML models and investigate the new Mel Frequency Cepstral Coefficients (MFCC) features. The obtained results show that our proposed methods can improve significantly the accuracy of these ML based models in recognizing and classifying the bee buzzing form other ambient noises, making them even better than plural deep learning algorithms suggested in the literature. In addition, we introduce a new dataset of bee sound samples and we verify the efficiency of our proposed models on the new dataset.
... Manual journaling from the beekeeper is often accompanied by automation systems that monitor the current data of an apiary including the beehives and the microclimate of the area. Monitored data is captured and analysed by a number of systems and techniques including: audio / acoustics or sound analysis [22][23][24][25], motion/track analysis [26][27][28][29], population estimation and variability [30][31][32][33][34], behaviour analysis [35], vibration [36,37], image analysis and computer vision for detecting diseases and parasites [38], energy consumption [39], environmental data [40] etc. A number of systems have also been developed to monitor combinations of the aforementioned data based on multi-sensor arrays that also fall within the domain of IOT. ...
Chapter
Full-text available
In this paper, we present an ongoing project that focuses on designing interactive systems and their respective interfaces for monitoring and journaling of apiculture information acquired in the actual field during apiary inspections by the beekeepers. Initially, the paper provides a brief overview of the concepts and technologies found in the domain. Next, it examines the scenarios to be used for the design of interactions related to the actual beehive inspections and desktop use in the office. The paper mainly focuses on the design requirements based on user research. It provides a review of interaction techniques that can be implemented for journaling in the workplace of the apiary, briefly outlines the infrastructure and gives a system overview at its current state of development. Finally, the paper discusses future work including, guidelines towards the development of the various system components for journaling and a preliminary evaluation plan for the case studies that will follow.KeywordsJournaling for beekeepingBeehive inspectionApicultureIoTInteraction designAgriculture scienceBee management practicesModern beekeeping
Article
Full-text available
Beekeepers have experienced significant losses to the population of their hives in recent years due to a phenomenon that was once referred to as the Colony Collapse Disorder (CCD). In the past couple of decades, there have been some efforts to determine the causes of this disaster and how they can be addressed. Some of the research has relied on the data that the beekeepers collect manually from their hives. Some of the more recent efforts have analyzed audio and video recordings obtained at the hives to better understand the behavior of bees and determine health status of the hives. Such research requires quality audio, video, and other sensor data captured using a reliable system. To make this practical, the system should be inexpensive and with minimal disruption to the bees’ natural behavior at their hives. Moreover, such a system should be capable of providing insights and warnings to facilitate early intervention and help mitigate the dire consequences. This paper provides details on the design and implementation of a data collection and monitoring system that was created in our research lab and is referred to here as Beemon. This system automatically captures sensor data (temperature, humidity, weight) and sends them using MQ Telemetry Transport (MQTT) protocol to a ThingsBoard dashboard. The system also sends captured video and audio recordings at the hives’ entrance to our remote server for analysis and further research. Beemon operates continuously in an outdoor apiary environment and allows a near real-time data collection. This paper present limited results of several years of real-world operations to demonstrate the purpose of the proposed Beemon system.
Article
Full-text available
Recent years have seen a worsening in the decline of honey bees (Apis mellifera L.) colonies. This phenomenon has sparked a great amount of attention regarding the need for intense bee hive monitoring, in order to identify possible causes, and design corresponding countermeasures. Honey bees have a key role in pollination services of both cultivated and spontaneous flora, and the increase in bee mortality could lead to an ecological and economical damage. Despite many smart monitoring systems for honey bees and bee hives, relying on different sensors and measured quantities, have been proposed over the years, the most promising ones are based on sound analysis. Sounds are used by the bees to communicate within the hive, and their analysis can reveal useful information to understand the colony health status and to detect sudden variations, just by using a simple microphone and an acquisition system. The work here presented aims to provide a review of the most interesting approaches proposed over the years for honey bees sound analysis and the type of knowledge about bees that can be extracted from sounds.
Article
Full-text available
In this work, we disclose a non-invasive method for the monitoring and predicting of the swarming process within honeybee colonies, using vibro-acoustic information. Two machine learning algorithms are presented for the prediction of swarming, based on vibration data recorded using accelerometers placed in the heart of honeybee hives. Both algorithms successfully discriminate between colonies intending and not intending to swarm with a high degree of accuracy, over 90% for each method, with successful swarming prediction up to 30 days prior to the event. We show that instantaneous vibrational spectra predict the swarming within the swarming season only, and that this limitation can be lifted provided that the history of the evolution of the spectra is accounted for. We also disclose queen toots and quacks, showing statistics of the occurrence of queen pipes over the entire swarming season. From this we were able to determine that (1) tooting always precedes quacking, (2) under natural conditions there is a 4 to 7 day period without queen tooting following the exit of the primary swarm, and (3) human intervention, such as queen clipping and the opening of a hive, causes strong interferences with important mechanisms for the prevention of simultaneous rival queen emergence.
Article
Full-text available
The widespread decline of honey bee (Apis mellifera L.) colonies registered in recent years has raised great attention to the need of gathering deeper knowledge about this phenomenon, by observing the colonies’ activity to identify possible causes, and design corresponding countermeasures. In fact, honey bees have well-known positive effects on both the environment and human life, and their preservation becomes critical not only for ecological reasons, but also for the social and economic development of rural communities. Smart sensor systems are being developed for real-time and long-term measurement of relevant parameters related to beehive conditions, such as the hive weight, sounds emitted by the bees, temperature, humidity, and CO 2 inside the beehive, as well as weather conditions outside. This paper presents a multisensor platform designed to measure the aforementioned parameters from beehives deployed in the field, and shows how the fusion of different sensor measurements may provide insights on the status of the colony, its interaction with the surrounding environment, and the influence of climatic conditions.
Article
Full-text available
Beekeeping is one of the widespread and traditional fields in agriculture, where Internet of Things (IoT)-based solutions and machine learning approaches can ease and improve beehive management significantly. A particularly important activity is bee swarming. A beehive monitoring system can be applied for digital farming to alert the user via a service about the beginning of swarming, which requires a response. An IoT-based bee activity acoustic classification system is proposed in this paper. The audio data needed for acoustic training was collected from the Open Source Beehives Project. The input audio signal was converted into feature vectors, using the Mel-Frequency Cepstral Coefficients (with cepstral mean normalization) and Linear Predictive Coding. The influence of the acoustic background noise and denoising procedure was evaluated in an additional step. Different Hidden Markov Models' and Gaussian Mixture Models' topologies were developed for acoustic modeling, with the objective being to determine the most suitable one for the proposed IoT-based solution. The evaluation was carried out with a separate test set, in order to successfully classify sound between the normal and swarming conditions in a beehive. The evaluation results showed that good acoustic classification performance can be achieved with the proposed system.
Article
Full-text available
Electronic beehive monitoring extracts critical information on colony behavior and phenology without invasive beehive inspections and transportation costs. As an integral component of electronic beehive monitoring, audio beehive monitoring has the potential to automate the identification of various stressors for honeybee colonies from beehive audio samples. In this investigation, we designed several convolutional neural networks and compared their performance with four standard machine learning methods (logistic regression, k-nearest neighbors, support vector machines, and random forests) in classifying audio samples from microphones deployed above landing pads of Langstroth beehives. On a dataset of 10,260 audio samples where the training and testing samples were separated from the validation samples by beehive and location, a shallower raw audio convolutional neural network with a custom layer outperformed three deeper raw audio convolutional neural networks without custom layers and performed on par with the four machine learning methods trained to classify feature vectors extracted from raw audio samples. On a more challenging dataset of 12,914 audio samples where the training and testing samples were separated from the validation samples by beehive, location, time, and bee race, all raw audio convolutional neural networks performed better than the four machine learning methods and a convolutional neural network trained to classify spectrogram images of audio samples. A trained raw audio convolutional neural network was successfully tested in situ on a low voltage Raspberry Pi computer, which indicates that convolutional neural networks can be added to a repertoire of in situ audio classification algorithms for electronic beehive monitoring. The main trade-off between deep learning and standard machine learning is between feature engineering and training time: while the convolutional neural networks required no feature engineering and generalized better on the second, more challenging dataset, they took considerably more time to train than the machine learning methods. To ensure the replicability of our findings and to provide performance benchmarks for interested research and citizen science communities, we have made public our source code and our curated datasets.
Conference Paper
Swarming of honeybees is one of the biggest problems for the hobbyist-beekeepers. This is a state when old queen bee leaves the hive with worker bees. The aim of this work is to find relation between sound and swarming. The data was collected during the late spring and early summer using microphone that was placed inside the hive. Using the Welch method, implemented in Python, it was proven the relation between the power and swarming.
Article
Honey bees are of vital importance to global crop production. Colony losses have reached historic levels in Europe and North America and are high in other parts of the world. The decline in honey bee health has resulted in the demand for novel mechanisms of monitoring colony health by beekeepers and researchers. Methods of monitoring bee health traditionally involve opening of the hive either for manual data collection or the use of invasive electronic monitors. This study evaluates a beehive activity monitor based on the Doppler radar principle as a tool for assessing honey bee colony health. The research was conducted during a two-year study (2017–2018). We discuss the development of a portable Doppler radar unit and three experiments conducted with the aim of showing its utility in monitoring colony health. First we determined the relationship between: (I) forager activity and colony health (r² = 0.433, P = 0.006), (II) Doppler unit root mean square (RMS) and forager activity (r² = 0.766, P = 0.013), and then (III) Doppler unit RMS and colony health measured as total sum of brood and worker population (r² = 0.731, P = 0.026). This small portable Doppler unit is solar powered and can be deployed in any apiary to provide beekeepers with a tool for tracking their colonies' health in real time.
Article
Mobile phones can be found almost everywhere across the globe, upholding a direct point-to-point connection between the device and the broadcast tower. The emission of radiofrequency electromagnetic fields (RF-EMF) puts the surrounding environment inevitably into contact with this radiation. We have therefore exposed honey bee queen larvae to the radiation of a common mobile phone device (GSM band at 900 MHz) during all stages of their pre-adult development including pupation. After 14 days of exposure, hatching of adult queens was assessed and mating success after further 11 days, respectively. Moreover, full colonies were established of five of the untreated and four of the treated queens to contrast population dynamics. We found that mobile phone radiation had significantly reduced the hatching ratio but not the mating success. If treated queens had successfully mated, colony development was not adversely affected. We provide evidence that mobile phone radiation may alter pupal development, once succeeded this point, no further impairment has manifested in adulthood. Our results are discussed against the background of long-lasting consequences for colony performance and the possible implication on periodic colony losses.