Using Environmental Data for IoT Device Energy Harvesting

Prediction

Mansour Alzahrani

, Alex S. Weddell

and Gary B. Wills

School of Electronics and Computer Science, University of Southampton, Southampton, U.K.

Keywords: IoT, Energy Harvesting, Solar Energy Harvesting, Environmental Data, Weather, Machine Learning.

Abstract: There has been significant innovation in the domain of Internet of Things (IoT) as nowadays wireless data

transmission is playing an essential role in various organizations like agriculture, defence, transportation, etc.

Batteries are the most common option to power wireless devices. However, using batteries to power IoT

devices has drawbacks including the cost and disruption of frequent battery replacement, and environmental

concerns about battery disposal. Solar energy harvesting is a promising solution for long-term operation

applications. However, solar energy harvesting varies drastically over location and time. Due to fluctuating

weather conditions and the environmental effects on PV surface condition, output could be reduced and

become insufficient. Environmental conditions including temperature, wind, solar irradiance, humidity, tilt

angle and the dust accumulated over time on the photovoltaic (PV) module surface affects the amount of

energy harvested. To address this issue, a novel solution is required to autonomously predict the harvested

energy and plan the IoT device tasks accordingly, to enhance its performance and lifetime. Using Machine

Learning (ML) algorithms could make it possible to predict how much energy can be harvested using weather

forecast data. This research is ongoing, and aims to apply ML algorithms on historical weather data including

environmental factors to generate solar energy predictions for IoT device energy budget planning.

1 INTRODUCTION

It is estimated by Arm that about 1 trillion devices

will be connected and become part of everyday life

by the year 2035 (Sparks, 2017). The Internet of

Things (IoT) has the potential to make our

surroundings, houses, and vehicles smarter and more

quantifiable (Loh, 2021). Moreover, IoT devices such

as sensors, lights, and meters provide data and

information to smart cities for collection and analysis.

This data can be used to improve services,

infrastructure, utilities, and other aspects of life

(Gyrard, 2018).

A sensory IoT node consists of sensors, a

processor, a transceiver, and a power supply.

Batteries are the most common source of energy used

to supply power to wireless IoT nodes, which can

either be recharged or replaced according to the

installation conditions (Tong, 2011). Batteries are

often the largest part of an IoT device and, in most

cases, will need to be replaced during the system

lifetime—however, the cost of replacing batteries is

https://orcid.org/ 0000-0002-0458-773X

https://orcid.org/ 0000-0002-6763-5460

https://orcid.org/ 0000-0001-5771-4088

frequently higher than the cost of the IoT device

itself. Owing to cost and environmental concerns,

some systems now use alternative energy supplies

which can exploit ambient energy (Khan, 2015).

Energy can be harvested from various renewable and

ambient energy resources, for example,

vibration/movement, wind, solar, heat, or radio

frequency waves (Garg, 2017). Powering IoT devices

through harvesting sustainable energy has proved to

be an effective solution, extending their operation

lifetime and simplifying their installation. Solar

energy has gained more attention due to the

availability of light in many applications, along with

its simplicity and low component cost. However,

solar energy harvesting depends on the site's

conditions, such as geographical location, weather,

solar irradiance, temperature, and solar angle. The

output power of PV panels is dependent on the solar

irradiance. However, PV efficiency is indirectly

related to the other parameters, including relative

humidity, dust accumulation, wind speed, and

temperature (Yang, 2014). Dust accumulation on the

Alzahrani, M., Weddell, A. and Wills, G.

Using Environmental Data for IoT Device Energy Harvesting Prediction.

DOI: 10.5220/0011069700003194

In Proceedings of the 7th International Conference on Internet of Things, Big Data and Security (IoTBDS 2022), pages 197-204

ISBN: 978-989-758-564-7; ISSN: 2184-4976

197

surface causes shading, and hence performance

degradation. A sequence of rain and sandstorms

affects the adhesion of dust on the surface. We need

to be able to understand the complex environmental

conditions which affect the solar energy harvested.

Recently, much work has been done on power

forecasting and analyses of PVs. Different models

have been used for predictive analyses. Statistical and

machine learning (ML) models may give an insight

into the features of data dependencies and illustrate

the importance of individual characteristics

(Bergonzini, 2009).

This research aims to investigate the feasibility of

using ML and environmental data for solar energy

prediction, to increase system performance. It will

focus on predicting the amount of energy available in

the future, based on historical data and future public

weather forecasts. This paper gives a brief background

to IoT devices and energy harvesting, focussing on the

environmental impact on solar energy harvesting and

discuses some prediction algorithms (section 2). Next,

model design (section 3) gives an overview of the

proposed solar energy prediction model components

and workflow, including a description of the data

acquisition needed for the model. Lastly, section 4

provides details of future work.

2 RESEARCH BACKGROUND

2.1 Wireless IoT Devices

Wireless smart sensors are becoming more critical in

the development of the Internet of Things. These

smart devices are aimed to measure and monitor

natural conditions like temperature, dampness, sound,

pressure, air quality (Newell, 2019). A major

complication for such smart IoT devices being

genuinely ubiquitous is their requirement for a long-

term, dependable power supply. Batteries are the most

common option to power wireless devices. Still, they

must be replaced regularly to guarantee continued

functioning. Such a requirement is unattractive since it

implies significant maintenance expenses, particularly

in distant locations. Energy harvesting from solar,

wind, thermal, and RF sources has been suggested to

solve these issues (Khan, 2015).

2.2 Solar Energy for Wireless IoT

Devices

Photovoltaic (PV) cells convert light energy to

electricity. Solar is one of the most promising and

prevalent forms of renewable energy that utilizes

either natural or artificial light to generate electricity

(Piñuela, 2013). Solar energy harvesting enables the

operation of IoT nodes sustainably and can simplify

their deployment. PV energy harvesting, unlike other

harvesting sources such as temperature difference,

vibration or airflow, is widely available in sufficient

quantities to make the powering of low-power IoT

devices practical.

As solar energy harvesting varies significantly

over time, the energy collected must sometimes be

stored to be utilized when the energy source dictates.

As a result, the Harvest-Store-Use method is well

suited to dealing with unstable energy sources

(Choudhary, 2020).

2.3 Environmental Impact on Solar

Energy Harvesting for Wireless IoT

Devices

Solar energy fluctuations are determined by seasonal

climatic and weather variables such as temperature,

hourly solar angle, solar irradiance, the orientation of

the solar panel and tilt angle and shadows (Yadav,

2014). The following sub-sections will explore the

literature on the impact of various environmental

factors on PV module performance.

2.3.1 Effect of Temperature on PV Module

Performance

Around 17-20% of solar energy is converted into

electricity by a PV module. However, some of the

wasted solar energy appears as heat. The increase in

temperature of the PV module has negative impact on

its efficiency. The electrical yield is reduced with an

increase in the temperature of the module (Rahman,

2017). According to D. Du, the efficiency of

crystalline silicon PV cell drops around 0.45%/°C

(Du, 2013). In an area with an irradiation level of

1000 W/m2, the PV temperature increased to around

56 °C, causing the module output power dropped to

from 49.89 W to 29.42 W and the electrical efficiency

decrease 3.13% (Rahman, 2015), furthermore at

around 25 °C ambient temperature is the maximum

PV efficiency can be achieved. The temperature of a

PV module can be calculated by this equation:

Tc=Tamb+(NOCT−20) G/800 (1)

Where,

Tc: Cell temperature

Tamb: ambient temperature

(NOCT): The normal operating cell temperature

G: Irradiance

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

198

2.3.2 Effect of Dust Accumulation and Tilt

Angle on PV Module Performance

The capability of the cover glass on the surface of the

solar cell module to transmit the solar light has a

significant impact on the solar panel's performance.

Over time, the transmittance may decrease due to dust

accumulation on the glass, which depends on the

collection of dust on the panel surface (Sarver, 2013).

After six months of deployment in the Kingdom of

Saudi Arabia (KSA) environment, PV module

harvested power decrease by 50% (Adinoyi, 2013);

furthermore, the power outputs decreased by 20%

after one sandstorm. The dust accumulation on the

surface of the solar PV panel causes the short circuit

current to drop off at a faster rate, mainly when the

density of dust is higher. There is a 1.7 % loss in PV

power per g/m2 when dust accumulates on the surface

of the PV. Both outdoor and indoor circumstances

were validated for this correlation (Dhaundiyal,

2020).

In dusty environments, the tilt angle also has an

impact on dust accumulation. The deposition of dust

particles on the PV surface with tilt angle was studied

by Sayigh et al. (Sayigh, 1985), who experimented in

Kuwait City. Exposing the panels outside for 38 days,

Sayigh and his team observed that 17% to 64% of

plate transmittance is reduced when the tilt angle is

changed from 60° to 0°, respectively. Another study

shows that, after exposing the PV panels outside for

14 days, results reported the efficiency dropped by

37.63%, 14.11%, and 10.95% for the 0°, 25°, and 45°

tilt angle, respectively (Hachicha, 2019).

2.3.3 Effect of Wind Speed and Direction on

PV Module Performance

The dissipation of convective heat transmission of the

solar module is enhanced by air flow, hence lowering

the temperature of the panel and helps to sustain the

conversion performance (Mazón, 2011). In Dhahran,

KSA, for example, the module's temperature is

reduced by around 10 °C by the increase in velocity

of wind from 2.8 to 5.3 m/s (Said, 2015). Several

experiments were held, artificially varying the wind

direction and velocity to examine the performance of

solar PV. Dust particles are blown away from the

surface of solar PV by the wind, diminishing dust

deposition. In Egypt, it is seen that after two weeks of

wind, the rate of dust deposition on a PV panel

surface decreased significantly (Hegazy, 2001).

However, the impact of wind direction is not

appropriately addressed. Wind coming from the

desert should be warmer than from the sea or

lakeside. Hence the direction wind also plays a

significant role in the temperature of the PV module.

2.3.4 Effect of Humidity on PV Module

Performance

It is observed that the solar PV efficiency is increased

when humidity is relatively low. The impact on the

performance of solar modules due to relative

humidity has been verified. The performance yield of

solar cells is enhanced from 9.7% to 12.04% at 60%

to 48% humidity, respectively (Katkar, 2011). As far

as power is concerned, it is noted that with the

increase in relative humidity by 20% in result

approximately 12.4% of power is decreased

(Rahman, 2015). Concerning dust adhesion and

humidity, it is noticed that dust particles stick on the

panel surface due to humidity. In order to restore the

module to its initial power efficiency, cleaning the PV

module's surface is required. In respect of quantitative

analysis, it is seen that adhesion is increased by

approximately 80% when relative humidity is

increased from 40% to 80% (Said, 2014).

2.3.5 Effect of Rainfall on PV Module

Performance

Rainfall has an impact on solar PV, as rain removes

dense dust from the panels. However, some particles

of dust stick on the panel surface due to cementation

and might not be detached. Michel and Muller

(Micheli, 2017) have shown the correlation between

surface cleaning and a rain event. However, a

minimum of precipitated water is required for

effective cleaning. Different reports provide the

minimum rain threshold required for cleaning the

panel, such as minimal of 5mm daily rainfall (García,

2011) and 6.9mm (Toth, 2020). The intricacy of the

surface cleaning can be attributable to the multiplicity

of threshold conditions because they are further

subject to different factors such as dust type,

wettability, speed of droplet, dust adhesion, and

surface inclination state (Ilse, 2018). In dry and semi-

dry regions, where the soiling rate is high and there is

less rainfall, rainfall is not considered sufficient for

dust removal. In such areas, a proper cleaning

mechanism is required for better performance.

2.3.6 Effect of Air Quality on PV Module

Performance

Air quality is an important parameter, since the

accumulation of ambient Particulate Matter (PM) on

the surface is the main reason for PV soiling.

Furthermore, soiling on the PV surface has an impact

Using Environmental Data for IoT Device Energy Harvesting Prediction

199

on the Direct Normal Irradiance (DNI); hence the PV

model performance will be decreased. PM with an

aerodynamic diameter of <10μm (PM

10) or <2.5 μm

(PM2.5 or PMfine) is the measurement of the air

quality. Micheli et al. found that ambient PM10

concentrations yielded better correlation in long dry

period than PM

2.5 concentration (Micheli, 2019). In

studies (Micheli, 2019) and (Coello, 2019), it has

been shown that PM concentrations variability are

important factors in modelling PV soiling. Moreover,

air quality has an impact on the horizontal visibility,

hence the Global Horizontal Irradiance (GHI)

reaching the PV surface is reduced by as much as 40%

to 50%, with a much more substantial reduction in the

Direct Normal Irradiance (DNI) at noon when the

Aerosol Optical Depth (AOD) is 3.0 (Kosmopoulos,

2017). Therefore, air quality is one of the parameters

that will be explored and used in developing a

predictive PV soiling model based on time-series.

2.4 Prediction Algorithms for Solar

Energy Harvesting

In the last couple of years, several prediction

techniques have been used to achieve the solar energy

prediction. These techniques can be distinguished

between Past Predicts the Future (PPF) and Weather

Forecast-Based Techniques (Sharma, 2010). PPF

techniques consider previously available data on how

much energy was harvested and apply that to the

future. These techniques divide the day into a number

of equal-sized time slots, and the prediction is done

either for the next slot based on the previous one or

for the same time slot of the next day. These are

simple techniques and easy to use but fail to give

valuable predictions when the weather changes.

Exponentially Weighted Moving Average

(EWMA) is an algorithm that was proposed by Kansall

et al; the algorithm has shown good predictions

accuracy results (Kansal, 2007). However, the EWMA

prediction accuracy is high only when weather

conditions are consistent. The reason for this is because

of the way the EWMA works. The EWMA algorithm

divides the day into a number of fixed time slots.

Moreover, it predicts the energy harvesting rate based

on the weighted average for that period in previous

days. When the weather changes frequently, and there

is a mix of sunny and cloudy periods, the EWMA

algorithm gives low accuracy predictions.

Piorno et al. proposed the Weather-Conditions

Moving Average (WCMA) algorithm to solve these

shortcomings of the EWMA (Piorno, 2009). It

includes a GAP factor, which is supposed to take into

account the average harvesting of previous days and

compare it to actual harvesting, thus determining the

weather direction change and considering it in the

prediction. While this does make better predictions

compared to EWMA, once again, when weather

changes are more frequent, and the fluctuations are

significantly larger, so are the predicting errors. These

predictions techniques are based on historical data,

i.e. how much energy was harvested in the past.

Nevertheless, this historical approach is unreliable

when weather changes occur between days and even

during a single day. To be able to make better energy

harvesting predictions and to overcome the

limitations of the PPF techniques, such as their short-

term focus, weather forecasts need to be included.

Weather forecasting has been used with historical

data to provide better predictions of future energy

harvesting. A study by (Sharma, 2010) revealed that

weather forecast-based predictions give better results

compared to PPF.

The Artificial Neural Network (ANN) and Support

Vector Machine (SVM) is widely used forecasting

techniques for forecasting the nonlinear time series

data. SVM used for the prediction of daily and monthly

global solar radiation in (Belaid, 2016) proved to

requires few simple parameters to get good accuracy.

The energy predictions can be enhanced through

techniques that allow several profiles to be combined.

The machine learning approach will be applied on

automatic learning and improve the prediction.

3 SOLAR ENERGY PREDICTION

MODEL BASED ON WEATHER

DATA FOR SOLAR ENERGY

HARVESTING WIRELESS IOT

NODES

A solar energy prediction model will allow IoT nodes

to schedule their duty cycles and tasks based on

predicted energy. The goal of the proposed model

(data-driven approach) is to explore the machine

learning algorithms and the associated feature

extraction for better solar energy predictions.

Knowing the solar energy budget ahead of time and

planning the sensor node tasks accordingly, will aid

the sustainable operation of the sensor node.

3.1 Overview of the Solar Energy

Prediction Model

The proposed model has multiple data inputs and

prediction models (See Figure 1).

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

200

Data Acquisition is the key to accurate power

predication. This subsection discusses and describes

the required data for machine learning which will

serve the proposed model in this research.

Historical Weather Data

The data collected is of two different types. The first

data is from King Abdullah City for Atomic and

Renewable Energy (K.A. CARE) for Jeddah City,

hourly data from June 2017 - June 2020. This data

consists of various variables such as air temperature,

global horizontal irradiance, wind direction,

moisture, and zenith angle, etc (See Table 1). The

second data set is from openweathermap.org for the

same period as the first data set. This data consists of

various variables such as visibility, clouds, dewpoint,

rain, time of sunrise, and sunset (See Table 2). The

two data sets will be combined and used to train and

test the prediction model. Dust Accumulation, Air

Quality, and Horizontal visibility are some of the

focuses of this research. Hence, they will be

investigated further, and the results added to the

model to enhance the prediction.

Figure 1: The Proposed Solar Energy Prediction Model.

Weather Forecast Data

The weather forecast data will be gathered from

openweathermap.org via their accessible API

(Openweathermap.org, 2022). This forecast data is

hourly. As part of the data preparation, the coldness

levels need to be changed to numerical values.

Photovoltaic Data

This data includes PV size, efficiency, peak power

point current and voltage, tilt angle, and orientation.

Energy Data from Sensor Node

- Battery Level: The battery level (state-of-charge,

SoC) can be estimated based on the battery voltage

measurement. The battery level will be presented as

a percentage from (0 - 100 %) based on the battery

voltage measurement and the energy storage

properties. This data will be gathered for each time

slot from the sensor node.

- Solar Energy Harvested: The amount of energy

harvested will be calculated based on the

measurement of the current from the solar panel,

considering the charging state. Knowing the charging

state is essential for the machine learning algorithms.

Sun Position and Tilt Angle

The sun position will be calculated hourly based on

the location of the sensor, time of the day, azimuth,

and zenith. The PV tilt angle and the orientation are

important for calculating the incidence angle based on

the sun's position. The amount of solar irradiance

reaching the surface of the PV can be determined by

the incidence angle and the size of the PV.

Prediction Models

In this work, widely used machine learning models

will be used for predictions, including k-nearest

neighbour (k-NN), support vector machines (SVM),

artificial neural networks (ANN). Different tests will

be conducted for fine tuning of each algorithm.

The accuracy of the models is dependent upon

different factors, which are as the following:

• The accurate measurement of weather forecast

comparing to the actual weather data.

• The missing data is either due to transmission

problems or the failure of nodes at any point in

time.

The following will explain each prediction in the

proposed model:

The Dust Accumulation Prediction: will

estimate the dust accumulation on the PV surface

over time based on the weather forecast data such as

horizontal visibility, air quality, wind speed, wind

directions, and rainfall considering the installation

date. This prediction will have direct influence on the

solar energy harvesting.

The Solar Energy Harvesting Prediction: This

prediction will be based on the historical weather

data, weather forecast data, PV data, the sun position

and tilt angle, and the energy from the sensor node

with respect of dust accumulation prediction.

Using Environmental Data for IoT Device Energy Harvesting Prediction

201

Table 1: list of the data Parameters and Description from (K.A. CARE).

Parameters Description

Air Temperature The degree of hotness or coldness of the environment (measured by C°)

Wind Direction at 3m Average wind direction at 3 meters height (measured by degree from North)

Wind Direction at 3m (std dev) Standard deviation of the average wind direction data at 3 meters height

Wind speed at 3m Average wind speed at 3 meters height (measured by m/s)

Wind speed at 3m (std dev) Standard deviation of the average wind speed data at 3 meters height

Azimuth Angle Defines the direction of the sun, Azimuth Angle is the angle between a line due south

and the shadow cast by a vertical rod on Earth (measured by degree)

Diffuse Horizontal Irradiance

(DHI)

Diffuse Horizontal Irradiance is the amount of radiation received per unit area by a

surface that does not arrive on a direct path from the sun, but has been scattered by

molecules and particles in the atmosphere (measured by W/m2)

Direct Normal Irradiance (DNI) Direct Normal Irradiance is the amount of solar radiation received per unit area by a

surface that is always held perpendicular to the rays that come in a straight line from

the direction of the sun (measured by W/m2)

Global Horizontal Irradiance

(GHI)

Global Horizontal Irradiance is the total amount of shortwave radiation received

from above by a surface horizontal to the ground. This value includes both Direct

Normal Irradiance and Diffuse Horizontal Irradiance (measured by W/m2)

Horizontal Visibility The greatest distance toward the horizon that prominent objects can be identified

visually with the naked eye (measured by km)

Peak Wind Direction at 3m Greatest value of wind direction at 3 meters height (measured by degree from North)

Peak Wind Speed at 3m Greatest value of wind speed at 3 meters height (measured by m/s)

Relative Humidity Relative humidity is the ratio of the partial pressure of water vapor to the equilibrium

vapor pressure of water at the same temperature (measured by %)

Barometric Pressure Atmospheric pressure or barometric pressure, is the pressure exerted by the weight

of air in the atmosphere of Earth (measured by mBar)

Zenith Angle The solar zenith angle is the angle between the zenith and the center of the sun

(measured by degree)

Table 2: Lists of The Data Required for Modelling.

DATA LIST

Historical Weather Data

Air Temperature, Diffuse Horizontal Irradiance (DHI), Direct Normal

Irradiance (DNI), Global Horizontal Irradiance (GHI), Azimuth Angle, Zenith

Angle, Horizontal Visibility, Peak Wind Direction at 3m, Peak Wind Speed at

3m, Wind Direction at 3m, Wind Direction at 3m (std dev), Wind speed at 3m,

Wind speed at 3m (std dev), Relative Humidity, Barometric Pressure, Time and

Date.

Weather Forecast and Alerts Wind Speed and Direction, Temperature, Pressure, Humidity, Air Quality,

Visibility, Clouds, Dewpoint, Rain, Time and Date.

Photovoltaic Data PV Size, Efficiency, Voltage at Peak Power, Current at Peak Power.

Sun Position and PV Tilt Angle Zenith, Azimuth.

Location, time of sunrise and sunset. PV Tilt Angle and Orientation.

Energy Data from Node Battery level, Solar energy harvested.

Battery Gain Prediction: Based on the

prediction of The Solar Energy Harvesting and the

battery level measurement at night when there is no

solar energy, taking into account the energy storage

discharging cycle. The battery level data and the

actual energy harvested data will indicate the energy

consumption for the sensor tasks operation. Hence,

this information will be used for more machine

learning to improve the predictions.

Battery Level Prediction: Based on the

prediction of the battery gain and the energy

consumption for the given tasks. The energy

consumption will be determined by how many tasks

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

202

will be scheduled for that day. The model will use the

data from the previous days to improve the prediction

accuracy over time.

Tasks Scheduling: When scheduling tasks for

IoT node, two factors need to be considered: the

amount of energy will be consumed in that time

period and the maximum utilization from the assigned

energy. Different Tasks Schedules will be tested to

Figure 2: Block diagram for the proposed Wireless IoT

Node Architecture.

4 CONCLUSIONS AND FUTURE

WORK

Wireless IoT sensors are widely used nowadays.

Sensors can be employed in remote locations and

harsh environments. Their operation lifetime depends

on their energy supply. Powering such devices via

solar energy harvesting systems enables the

continuous work. For

optimal use of the available

energy, the IoT device can schedule its operation

according to the available energy. In order to achieve

this, the prediction of the future energy which can be

harvested is required. Therefore, understanding the

effects of the environment on the solar energy

harvesting needed.

This research will be defining the most relevant

weather parameters and scenarios based on the

literature. Identify the possible scenarios help in

assessing the model in several operating conditions.

The developed scenarios will be simulated to

understand the effect of these scenarios on energy

harvesting. For example, if it rains after a dusty day,

what is the impact on solar energy harvesting.

Furthermore, the effective input features for the

prediction will be studied and identified and will be

tested in the simulation phase. The widely used

machine learning algorithms Support Vector

Machines (SVM) is most known as state-of-the-arts

forecasting models based on machine learning. These

models are data-driven, and they are suitable for

short-term as intra-hours and long-term as next-day

forecasts. These forecasting models will be tested in

terms of prediction accuracy and energy efficiency.

REFERENCES

Adinoyi, M. J., & Said, S. A. (2013). Effect of dust

accumulation on the power outputs of solar

photovoltaic modules. Renewable energy, 60, 633-636.

Belaid, S., & Mellit, A. (2016). Prediction of daily and

mean monthly global solar radiation using support

vector machine in an arid climate. Energy Conversion

and Management, 118, 105-118.

Bergonzini, C., Brunelli, D., & Benini, L. (2009, June).

Algorithms for harvested energy prediction in

batteryless wireless sensor networks. In 2009 3rd

International workshop on advances in sensors and

interfaces (pp. 144-149). IEEE.

Choudhary, P., Bhargava, L., Singh, V., Choudhary, M., &

kumar Suhag, A. (2020). A survey–Energy harvesting

sources and techniques for internet of things devices.

Materials Today: Proceedings, 30, 52-56.

Coello, M., & Boyle, L. (2019). Simple model for

predicting time series soiling of photovoltaic panels.

IEEE J. Photovolt, 9(5), 1382-1387.

Dhaundiyal, A., & Atsu, D. (2020). The effect of wind on

the temperature distribution of photovoltaic modules.

Solar Energy, 201, 259-267.

Du, D., Darkwa, J., & Kokogiannakis, G. (2013). Thermal

management systems for photovoltaics (PV)

installations: a critical review. Solar Energy, 97, 238-

254.

García, M., Marroyo, L., Lorenzo, E., & Pérez, M. (2011).

Soiling and other optical losses in solar‐tracking PV

plants in navarra. Progress in Photovoltaics: Research

and Applications, 19(2), 211-217.

Garg, N., & Garg, R. (2017, December). Energy harvesting

in IoT devices: A survey. In 2017 International

Conference on Intelligent Sustainable Systems (ICISS)

(pp. 127-131). IEEE.

Gyrard, A., Zimmermann, A., & Sheth, A. (2018). Building

IoT-based applications for smart cities: How can

ontology catalogs help?. IEEE Internet Things J, 5(5),

3978-3990.

Hachicha, A. A., Al-Sawafta, I., & Said, Z. (2019). Impact

of dust on the performance of solar photovoltaic (PV)

systems under United Arab Emirates weather

conditions. Renewable Energy, 141, 287-297.

Hegazy, A. A. (2001). Effect of dust accumulation on solar

transmittance through glass covers of plate-type

collectors. Renewable energy, 22(4), 525-540.

Ilse, K. K., Figgis, B. W., Naumann, V., Hagendorf, C., &

Bagdahn, J. (2018). Fundamentals of soiling processes

on photovoltaic modules. Renewable and Sustainable

Energy Reviews, 98, 239-254.

Kansal, A., Hsu, J., Zahedi, S., & Srivastava, M. B. (2007).

Power management in energy harvesting sensor

Using Environmental Data for IoT Device Energy Harvesting Prediction

203

networks. ACM Trans. Embedded Comput Syst

(TECS), 6(4), 32-es.

Katkar, A. A., Shinde, N. N., & Patil, P. S. (2011).

Performance & evaluation of industrial solar cell wrt

temperature and humidity. International Journal of

research In Mechanical engineering and technology,

1(1), 69-73.

Khan, J. A., Qureshi, H. K., & Iqbal, A. (2015). Energy

management in wireless sensor networks: A survey.

Computers & Electrical Engineering, 41, 159-176.

Kosmopoulos, P. G., Kazadzis, S., Taylor, M.,

Athanasopoulou, E., Speyer, O., Raptis, P. I., &

Kontoes, C. (2017). Dust impact on surface solar

irradiance assessed with model simulations, satellite

observations and ground-based measurements.

Atmospheric Measurement Techniques, 10(7), 2435.

Loh, F., Geißler, S., Schaible, F., & Hoßfeld, T. (2021,

January). Talk to Me: Investigating the Traffic

Characteristics of Amazon Echo Dot and Google

Home. In 2020 IEEE Eighth International Conference

on Communications and Electronics (ICCE) (pp. 521-

526). IEEE

Mazón, R., Káiser, A. S., Zamora, B., García, J. R., & Vera,

F. (2011, November). Analytical model and

experimental validation of the heat transfer and the

induced flow in a PV cooling duct in environmental

conditions. In World Renewable Energy Congress-

Sweden; 8-13 May; 2011; (No. 057, pp. 2907-2915).

Micheli, L., & Muller, M. (2017). An investigation of the

key parameters for predicting PV soiling losses.

Progress in photovoltaics: research and applications,

25(4), 291-307.

Micheli, L., Deceglie, M. G., & Muller, M. (2019).

Predicting photovoltaic soiling losses using

environmental parameters: An update. Progress in

Photovoltaics: Research and Applications, 27(3), 210-

219.

Newell, D., & Duffy, M. (2019). Review of power

conversion and energy management for low-power,

low-voltage energy harvesting powered wireless

sensors. IEEE Trans. Power Electron, 34(10), 9794.

Piñuela, M., Mitcheson, P. D., & Lucyszyn, S. (2013).

Ambient RF energy harvesting in urban and semi-urban

environments. IEEE Trans. Microw. Theory techn,

61(7), 2715-2726.

Piorno, J. R., Bergonzini, C., Atienza, D., & Rosing, T. S.

(2009, May). Prediction and management in energy

harvested wireless sensor nodes. In 2009 1st

International Conference on Wireless Communication,

Vehicular Technology, Information Theory and

Aerosp. Electron Syst Technol (pp. 6-10). IEEE.

Rahman, M. M., Hasanuzzaman, M., & Abd Rahim, N.

(2017). Effects of operational conditions on the energy

efficiency of photovoltaic modules operating in

Malaysia. Journal of cleaner production, 143, 912-924.

Rahman, M. M., Hasanuzzaman, M., & Rahim, N. A.

(2015). Effects of various parameters on PV-module

power and efficiency. Energy Conversion and

Management, 103, 348-358.

Said, S. A., & Walwil, H. M. (2014). Fundamental studies

on dust fouling effects on PV module performance.

Solar Energy, 107, 328-337.

Said, S. A., Al-Aqeeli, N., & Walwil, H. M. (2015). The

potential of using textured and anti-reflective coated

glasses in minimizing dust fouling. Solar Energy, 113,

295-302.

Sarver, T., Al-Qaraghuli, A., & Kazmerski, L. L. (2013). A

comprehensive review of the impact of dust on the use

of solar energy: History, investigations, results,

literature, and mitigation approaches. Renewable and

sustainable energy Reviews, 22, 698-733.

Sayigh, A., Al-Jandal, S., & Ahmed, H. (1985, September).

Dust effect on solar flat surfaces devices in Kuwait. In

Proceedings of the workshop on the physics of non-

conventional energy sources and materials science for

energy (pp. 353-367). ICTP Triest, Italy.

Sharma, N., Gummeson, J., Irwin, D., & Shenoy, P. (2010,

June). Cloudy computing: Leveraging weather

forecasts in energy harvesting sensor systems. In 2010

7th Annual IEEE Communications Society Conference

on Sensor, Mesh and Ad Hoc Communications and

Networks (SECON) (pp. 1-9). IEEE.

Sparks, P. (2017). The route to a trillion devices. White

Paper, ARM.

Tong, B., Wang, G., Zhang, W., & Wang, C. (2011). Node

reclamation and replacement for long-lived sensor

networks. IEEE Trans. Parallel Distrib. Syst, 22(9),

1550-1563.

Toth, S., Hannigan, M., Vance, M., & Deceglie, M. (2020).

Predicting photovoltaic soiling from air quality

measurements. IEEE J. Photovolt, 10(4), 1142-1147.

Weather API. (n.d.). Retrieved 11 January 2022, from

Openweathermap.org website: https://openweather

map.org/api

Yadav, A. K., & Chandel, S. S. (2014). Solar radiation

prediction using Artificial Neural Network techniques:

A review. Renewable and sustainable energy reviews,

33, 772-781.

Yang, H. T., Huang, C. M., Huang, Y. C., & Pai, Y. S.

(2014). A weather-based hybrid method for 1-day

ahead hourly forecasting of PV power output. IEEE

trans. Sustain.Energy, 5(3), 917-92.

IoTBDS 2022 - 7th International Conference on Internet of Things, Big Data and Security

204