Theoretical model and implementation of a real time intelligent bin ...

Expert Systems With Applications 48 (2016) 76–88

Contents lists available at ScienceDirect

Expert Systems With Applications journal homepage: www.elsevier.com/locate/eswa

Theoretical model and implementation of a real time intelligent bin status monitoring system using rule based decision algorithms Md. Abdulla Al Mamun a,∗, M A Hannan a, Aini Hussain a, Hassan Basri b a Department of Electrical, Electronic and Systems Engineering, Faculty of Engineering & Built Environment, Universiti Kebangsaan Malaysia, Bangi, Selangor DE, Malaysia b Department of Civil and Structural Engineering, Faculty of Engineering & Built Environment, Universiti Kebangsaan Malaysia, Bangi, Selangor DE, Malaysia

a r t i c l e

i n f o

Keywords: Solid waste Smart bin Decision algorithm Wireless sensor network ZigBee

a b s t r a c t Due to the rising trend of urbanization along with overconsumption of non-recyclable resources, the volume of municipal solid waste is increasing every day. An efficient, cost effective and environment friendly solution for real time bin status monitoring, collection and transportation of municipal solid waste is still a major challenge to the local municipal authorities. This research proposes a novel model, architecture and intelligent sensing algorithm for real time solid waste bin monitoring system that would contribute to the solid waste collection optimization. The monitoring application is based on decision algorithms for sensing solid waste data in a wireless sensor network. The system is built on a three level architecture like smart bin, gateway and control station. The elementary concept is that, smart bins collect their status when any changes occur and transmit the status data to a server via an intermediate coordinator. A set of applications in server presents the updated bin status on real time. The field test performances show that the system can efficiently monitor real time bin status that makes it feasible to decide, which bin should collect and which should not. Thus the proposed system has achieved its goal to provide real time bin status information to the solid waste management operator. Later, this information can be used for collection route optimization to reduce collection costs and carbon emissions which in turn contribute to build green society. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction The term Municipal Solid Waste (MSW), also named as garbage or trash, generally comprises of daily stuffs that we used and then thrown away. These stuffs include food scraps product packaging, newspapers, clothing, grass clippings, bottles, furniture, batteries, paint, appliances etc. that comes from our households, institutes, markets, hospitals, and trades etc. (Municipal Solid Waste, 2012). Due to the overconsumption of non-renewable resources, the volumes of MSW are increasing day by day. The generation as well as the recycling, composting, and disposal of MSW have changed substantially over the last few decades. In urban areas the waste generation rate is about 760,000 tons on daily basis and is anticipated to rise about 1.8 million tons per day by the year 2025 (World Bank, 1999). At present, the issues of waste collection and transport and its impact to human health due to pollutant emissions, noise, traffic, and so on is of big concern. The increasing numbers and unplanned usage of waste collection vehicles consumes a lot of fuel that turn in salient ∗

Corresponding author. Tel.: +880-1718890345. E-mail addresses: [email protected], [email protected], [email protected] (Md.A.A. Mamun), [email protected] (M.A. Hannan), [email protected] (A. Hussain), [email protected] (H. Basri). http://dx.doi.org/10.1016/j.eswa.2015.11.025 0957-4174/© 2015 Elsevier Ltd. All rights reserved.

contribution to gassy pollutant and greenhouse gas (GhG) emissions, mainly in city areas. The bins that are partially fill up when collecting seems an unnecessary wastage of resources. By optimizing the number of bin, location of the bin and frequency of their collection is an important way to reduce the cost and emission in solid waste collection (Badran & El-Haggar, 2006; Chang & Wei, 1999; Faccio, Persona, & Zanin, 2011; Johansson, 2006; Kulcar, 1996; Lin, Chen, Lee, & Lin, 2010). A waste collection truck with diesel engine emits an average of 2.4 kg/km for CO2 , 0.21 g/km for HC, 7.4 g/km for CO, 32.3 g/km for NOx and 46.4 mg/km for Particular Matter. And this data for a truck with CNG engine is 3.6 kg/km for CO2 , 2.19 g/km for HC, 15.8 g/km for CO, 4.38 g/km for NOx and 11.4 mg/km for Particular Matter (Fontaras et al., 2012). One reckoning presented that 5.87 L of fuel is consumed for every ton of waste collection which emits 4.40 kg CE of GHG ( Chen & Lin, 2008). Again, it is very expensive for the municipality to collect, transfer and transport solid waste. These operations constitute about 80–95% of the whole budget for solid waste management; so it figures the key element in ascertaining the finances of the entire waste management process (Alagöz & Kocasoy, 2008). Over the past few years, wireless sensor networks (WSN) have been deployed in various applications especially in case of remote monitoring, aiming to eliminate the web of wires as well as to reduce cost while extending network coverage. In the last decade, a

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

huge number of applications like habitat monitoring, building automation, smart energy, health care, water and air quality monitoring, construction health monitoring, agriculture and food industry, fire detection that had been developed in adopting ZigBee and GSM technology (Fraile, Bajo, Corchado, & Abraham, 2010; Han & Lim, 2010; Khedo, Perseedoss, & Mungur, 2010; Kistler, Bieri, Wettstein, & Klapproth, 2009; Mainwaring, Culler, Polastre, Szewczyk, & Anderson, 2002; Wang, Zhang, & Wang, 2006; Yu, Wang, & Meng, 2005). Now it is more awaited that WSN with currently developed more advance communication technologies leads to solve various problems towards building smart world and also brings significant benefits in reducing cost. The aim of this work is to design a frame work for collecting bin status data in real time that can help to optimize waste collection route by using the collected data to reduce operation costs and GhG emissions as well. 2. Related works In 1961, Solleftea hospital in Sweden installed the world’s first ever automatic refuse collection system named Centralsug (current Envac) which was pneumatic and then in 1965, the residential district of Or-Hallonbergen installed the first vacuum system for household waste management (Vacuum system history, n.d.). Still today, these systems are performing operations using the basic functions and structures from the early 1960 s. From that time, many researchers give contributions in the field of waste management, waste monitoring, incineration management, waste to energy conversion, and land filling (Arebey, Hannan, Begum, & Basri, 2012; Chen & Li, 2010; Hannan, Arebey, Begum, & Basri, 2011). Many researches have been undertaken to design and develop a system that can monitor and manage the solid waste collection process. (Cheng, Chan, & Huang, 2003; Lei, Chuanhua, Yuezhao, Haijun, & Yanghuiqin, 2011; Noche, Rhoma, Chinakupt, & Jawale, 2010). However, very little of research is done in the field of solid waste bin monitoring, specifically, on real time monitoring. An initiative was taken by the Swedish producers association, which equipped 3300 bins around the country with sensors and wireless communication equipment in order to estimate the bin fill level (Johansson, 2006). Each bin contains four infrared LED and a tilt sensor, mounted under the bin cover. The sensing system is activated once every hour and measures the fill level of the container. If three out of the four infrared beams are broken, the system is triggered and sent an alarm along with an email to the operator via GSM. A second alarm is sent in the same way when all the four beams are broken. It also sent a reset signal after the bin has been emptied (Johansson, 2006). The strengths of the system are that it can estimate bin fill level and facilitates to implement route optimization with low operation cost. The weaknesses of the system are the inability to measure exact fill level and weight along with delayed system responses. One research have been conducted to develop bin status monitoring system for the municipality of Pudong (Shanghai), by combining camera with others ICTs that can measure bin fill level as well as weight of waste inside bin in real time (Rovetta et al., 2009; Vicentini et al., 2009). As described by the authors, a camera is attached with a set of sensors such as ultrasonic, LEDs on the top of the bin, which enables it to collect information about the shape, area and height of the waste. The LEDs provide illumination to enable more accurate volume estimation. The bottom of the bin is equipped with a sensor which constantly scales the weight of the waste. GPRS module is installed with the bin to transmit acquired data to the control station. The functionalities of the system is that, bin status monitoring mainly focus on early gathering of data about the waste in the bins and secondly, the transmission of the information to the operation center software which maps, monitors and plans for route optimization with the help of GPS and GIS. The strengths of these studies are the enhancement of the bins with a variety of sensors and cameras.

77

However, the designed systems does not support a wireless sensor network for further fusion of the sensor data neither supports RFIDs for bin tagging and identification and GPRS in every bins increase the operation costs. Also the camera produces low quality images and estimates the level of the bin wrongly if the bin is dirty and position of the camera affects the system performance (Islam et al., 2014). Another system has been developed for bin status monitoring using various ICTs such volumetric sensor, RFID, weighing system, GPRS and GPS (Faccio et al., 2011). The system is designed with RFID tag at the bin level for identification; the RFID reader attached on the vehicle’s bin hook to keep the scanning distance below 1 m; a programmable microprocessor is installed inside the bin to manage the detection measures of fill level using a volumetric ultrasonic detection sensor; weighing system in the vehicle; GPRS module in the bins, vehicles and control center; and software applications in vehicle and control center to trace next bins to be served and to collect and analyze data accordingly. The strengths of the system are that it can estimate bin fill level and facilitates to implement route optimization with real time system responses. But the developed system has a high operation cost as every bin contain GPRS module and it is unable to measure weight. Arebey, Hannan, Basri, Begum, and Abdullah (2011); Arebey et al. (2012); Arebey, Hannan, and Basri (2013) and Hannan et al. (2011) studied that, the integration of a number of ICTs can estimate the quantity of waste as well as monitor trash bins and collection vehicles. They have developed a bin monitoring system with RFID, camera, GPS, GIS and GPRS. The system starts its operation with the driver being assigned with a specific vehicle and a specific route. The driver turns on the black box controller installed in the vehicle, which activates the RFID reader, the camera, the GPS and the GSM/GPRS modules in order to prepare the vehicle for transmitting information to a control station. The system is based on the wireless communication between the bins and vehicles, and between the vehicle and the control station. When the vehicle approaches the bin area, the RFID reader identifies the bin tag and the camera snaps two pictures before and after the collection to estimate the amount of waste. According to the authors, it is important to highlight that the collection operators are responsible for the adjustment of the camera to find the best direction for taking a proper image of the bin and its surroundings. Moreover, they need to open the lid for capturing the two images. The fill level of a bin is estimated based on a comparison of the images by using some image analysis procedures at the server. The strengths of the paper are the intelligent system incorporated for bins and trucks monitoring as well as the enhanced communication technologies. However, the model exploits data produced only from a specific type of data acquisition device and not considers real time bin information. Also the system has performance problems due to the placement of camera during bin status data collection. In Muthukumaran and Sarkar (2013) the authors have proposed a solid waste disposal system using mobile adhoc networks. The paper presents a model for waste collection of bins, distributed in a highly densely populated city in India. It is formed a dynamic multi-hop network that can provide real time information to municipal authorities. The system is able to monitor online and visualize the status of the bins for further use; due to a capacity sensor and adhoc transceivers embedded in the bins. The strengths of the paper are the incorporation of a dynamic multi-hop network along with the online monitoring and visualization utilities. However, the paper does not use a variety of sensors since the data produced only from a specific type of capacity sensor. In McLeod et al. (2014) the authors have proposed a model for remote monitoring of charity assets in order to improve collection efficiency. It is proposed a model from a major UK charity, in order to monitor bank and shop servicing requirements. The system incorporates sensors embedded into bins and uses tabu search methods; to develop dynamic scheduling and routing models for waste

78

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

collection. The paper uses capacity and pressure sensors for solid waste collection. The use of wireless sensor networks is implied. The strengths of the paper are the treatment of waste collection problem from the collection cost perspective along with the incorporation of tabu search methods. However, this system increases operational cost due to GPRS module in every bin. It is also abandonment to measure weight. One research has been conducted that used wireless sensor network for bin status monitoring (Catania & Ventura, 2014). The system used proximity and weight sensor installed inside a waste bin to monitor its status. As described, the system frequently measure its fill level and weight value and transmit the collected data to a coordinator installed in a nearby lamppost using IEEE 802.15.4. The coordinator forward the data to a collection point through GPRS/Wi-Fi. The authorities in the collection point take decision about a bin to unload based on the collected status information. The strengths of the system are that, it has a wireless sensor network and can monitor both the fill level and weight of a bin. But, distance between a bin and a coordinator is low which cause an increase in cost as more gateway is needed. Also, the system response is not real time as it is not intelligent enough. In Reis, Caetano, Pitarma, and Gonçalves (2015) and Reis, Pitarma, Goncalves, and Caetano (2014), the authors developed a system to facilitate the recycling process by monitoring the waste bins. The system comprised with RFID tag embedded in rubbish bag, pressure sensor installed in bin and remote server. The recycling center identifies and weighs each rubbish bag during waste disposal and the accumulated data is sent to the remote server using ZigBee. The strength of the system is that, it is a real time system developed using low cost resource limited devices. However, no measurement of bin fill level and suitable only for recyclable waste enclosed in rubbish bag with RFID tag are the weaknesses of the system. One research has been conducted to implement Pay As You Throw (PAYT) service by monitoring the weight of waste when an user throws waste inside a bin (Aravossis, Nikolaidou, & Fountzoula, 2015). The developed system used RFID tag for each customer while RFID reader, weight and volume scale are installed in each waste bin. As described by the authors, the system implemented a customized user identification process that provides local authorities and users with a virtual identity. The digital scale determines weight of the recyclables waste and transmits the output to a remote server. A web-based application facilitates both the parties about their current status. Low cost resource limited system is the strength of the study. The weakness of the system is the lacking of measurement of bin fill level. Also it is a non-real time system designed only for PAYT service. In Asimakopoulos et al. (2015) and Papalambrou, Karadimas, Gialelis, and Voyiatzis (2015), the authors developed a system to collect bin fill level information along with their identity. The system have ultrasonic sensor and active RFID tag installed on the bins called field unit and RFID reader installed in any approved personnel or installed in a vehicle of existing organized transportation systems called mobile sink. The active RFID tags in the field unit transmit periodically their identity and a sensed value regarding its fill level to the mobile sink. Finally, the mobile sink forward this information to an upper data and knowledge management system by using Bluetooth and Wi-Fi communication technology for further processing. Low cost resource limited system is the strength of this study. But the system response is partially real time and only considers fill level data to represent the bin status. Also the designed system does not support a wireless sensor network for further fusion of the sensor data neither supports RFIDs for bin tagging and identification. Table 1 presents a summary of solid waste monitoring and management systems developed over the last and current decades. The comparison is done based on used sensors and communication technologies, system response types after throwing waste inside bin and consideration of route planning by using the collected bin

information. Most of the systems used RFID, GPRS and GPS to monitor solid waste bins and collection vehicle but are limited to the detection of bin condition without serving other facilities and solving various problems related to waste management. The above mentioned systems have several limitations that raise barrier to implement an efficient and intelligent solid waste management system. Such limitations include partial system design (Arebey et al., 2012; Faccio et al., 2011; Hannan et al. 2011; Vicentini et al., 2009), lacking enough data (Arebey et al., 2012; Asimakopoulos et al., 2015; Hannan et al., 2011), expensive network structure (Faccio et al., 2011; Vicentini et al., 2009), deficiency of real time bin status information operators (Aravossis et al., 2015; Arebey et al., 2012; Hannan et al., 2011) etc. In this study, an automated bin status monitoring system is proposed to overcome all the weaknesses. In the proposed system, several sensing and communication technologies are employed and integrated to collect a bin status information considering the ambiance condition. To establish the proposed system as an intelligent and expert system, a set of rule based decision algorithms are designed to acquire various parameters about bin condition immediately after each waste throwing operation. Communication technologies are chosen which can cover maximum distance while spend low operational cost. The proposed system is able to provide real time bin status information which can be feed to other applications for implementing dynamic route planning and PAYT service. 3. Intelligent bin monitoring system The methodology of the intelligent bin monitoring system is split into three sections such as theoretical framework, physical architecture and functional procedure. The system is contrived with a big pile of intelligence procedure to support the power of the prototype and also justify its capability. Sections of the methodology are described in detail as follows: 3.1. Theoretical framework The theoretical framework of the real time bin monitoring system is based on the measurements of parameters related to bin status, data communication between sensor node and node coordinator and between node coordinator and server. An accelerometer sensor continuously keeps monitored when a bin cover is opening i.e. it keeps track on the magnitude of the raw data of the x, y and z axes of the accelerometer. If any change happens, then it measures the acceleration A.

A = ( Ax , A y , A z )

(1)

where Ax , Ay and Az are the acceleration towards x, y and z axis. An accelerometer sensor converts this physical quantity to an electrical signal and outputs voltages corresponding to the magnitudes of the accelerations. For a capacitive accelerometer, the relationship between the acceleration and the output voltage is calculated from Lyshevski (2002) and Andrejašic (2008) as follows.

Vx = a ×

V0 × m ks × d

(2)

where, Vx is the output voltage due to a displacement x in any axis, a is the acceleration, V0 is the input voltage, m, d and ks are constant related to the proof mass, electrodes distance and material coefficient of the capacitive accelerometer. When the proof mass is static, this means no acceleration (a = 0), and therefore, the output voltage is zero. During the opening of the bin cover there produce a movement in the proof mass which in turn accelerate the sensor (a > 0) and an output voltage Vx is produced proportional to the acceleration. A Hall effect sensor tracks a bin cover status by measuring the proximity between a conductor and a magnet placed on the upper edge of the bin and bottom edge of the cover. With the presence of an

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

79

Table 1 A summary of systems that used different sensors and communication technologies to monitor and manage solid waste over the last years. Bin level measurement

Level measurement method

Weight measurement

Communication technologies

System response

Planning

Lee and Thomas (2004) Isoaho and Peltoniemi (2004) Friedlos (2005)

×

–

×

VHFR, GPS

Instantaneous

×



Hydraulic pressure

×

GPS, GSM

Instantaneous



×

–

×

Delayed

×

Johansson (2006) O’Connor (2007)

 ×

Infrared LED –

× ×

Delayed Delayed

 

Chowdhury and Chowdhury (2007) Kietzmann (2008) Wilson and Vincent (2008) O’Connor (2008) Swedberg (2009) Rovetta et al. (2009), Vicentini et al. (2009) Nielsen, Ming, and Nielsen (2010) Chen and Li (2010)

×

–



RFID, Bluetooth, GPRS GSM/GPRS RFID, Bluetooth, GPRS RFID, Wi-Fi

Delayed

×

× ×

– –

× 

RFID, GSM/GPRS GPRS, GPS

Delayed Instantaneous

× ×

× × 

– – Ultrasonic ranger

 × 

RFID, GPS, GPRS RFID, GPRS, GPS GPRS, GPS

Delayed Delayed Instantaneous

 × 

×

–

×

RFID, GPS

Delayed





Infrared

×

Instantaneous

×



Image processing

×

Delayed

×

 

Volumetric Image processing

 ×

Optical Fiber/GPRS, Bluetooth/ IEEE802.15.4, GPS, RFID RFID, GPS, GSM/GPRS RFID, GPRS, GPS RFID, GPS, GSM/GPRS

Instantaneous Delayed

 ×



–

×

Instantaneous

×

 

Infrared Infrared

× ×



Capacity sensor

 

Capacity and pressure sensor Proximity sensor

×

–

× 

Authors

Arebey et al., 2011 (2012, 2013), Faccio et al. (2011) Hannan, Arebey, Begum, and Basri (2012) Longhi et al. (2012) McLeod et al. (2013) Lu, Chang, and Liao (2013) Muthukumaran and Sarkar (2013) McLeod et al. (2014) Catania and Ventura (2014) Reis et al. (2014), (2015) Aravossis et al. (2015) Asimakopoulos et al. (2015), Papalambrou et al. (2015)

Delayed Instantaneous

 ×

×

IEEE802.15.4, GSM/GPRS GSM Coaxial/Wi-Fi, Infrared, RFID, GPS –

Delayed



×

GSM

Delayed





Delayed





IEEE 802.15.4, GPRS/Wi-Fi ZigBee

Instantaneous

×

–



RFID

Delayed

×

Ultrasonic sensor

×

RFID, Bluetooth, Wi-Fi

Delayed

×

input voltage, a Lorentz force is created which produce a voltage by using the following equation (Ramsden, 2011).

VH = I × B ×

1 q0 × N × d

(3)

where, VH is the Hall voltage, I is the current passing through the conductor, B is the magnetic field perpendicular to the current, q0 is the magnitude of the charge carriers in the conductor, N is the number of charge carriers per unit volume and d is the thickness of the conductor. If the Hall voltage is measured with the absence of magnetic field, then the output is zero i.e. when the bin cover is open the output will zero or will produce a nominal voltage. Again, when the cover is closed, with the presence of magnet, the transducer produces a voltage proportional to the current and the magnetic field. With a known value of the magnetic field strength, the gap of the cover from the bin can be estimated which will be useful to decide whether the bin is overflowed. An ultrasonic level sensor measures the time-of-flight (Parvis & Carullo, 2001) i.e. the complete return trip time, an ultrasonic pulse

takes to transmit and receive its reflected echo between the sensor and the sensed material level as follows.

t = 2 × d/v

(4)

where t is the time-of-flight, d is the distance travelled by the wave and v is the propagation velocity. The sensor produces an output voltage proportional to the travelled distance that can be estimated by the following equation:

V0 = analog scale factor × distance VCC 2n

(5)

where, analog scale factor = and VCC is the reference voltage and n is the bit factor of the ultrasonic sensor. A load cell sensor based on a Wheatstone Bridge Network uses four strain gauges with four separate resistors. The resistance of the electrical conductor changes with the changes in length due to stress and it is virtually proportional to the applied strain. In the Wheatstone Bridge, voltage difference is calculated in two junctions by applying an excitation voltage to other two junctions

80

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

Fig. 1. Architecture of the real time bin status monitoring system.

(Huddleston, 2006) as follows.

V0 = Vi



R2 R1 − R1 + R4 R2 + R3

 (6)

where, V0 is the output voltage, V0 is the excitation voltage and R1 , R2 , R3 , R4 are the four registers. When a bin is empty, the voltage output will be zero i.e. the value of the four resistors are nearly same. Waste inside the bin causes a variation in value of one or more resistors due to the generated strain from the metallic member that contains the strain gauges. The output voltage is changed with this variation in resistance that is proportional to the weight of the waste. A temperature sensor with internal diode produces a change in output voltage proportional to the ambient temperature based on a temperature coefficient and known output voltage at 0 °C as follows (Khan, 1985).

V0 = ( TC × TA ) + V0 0 C

(7)

where, V0 is the sensor output voltage, V0 0 C is the sensor output voltage at0 °C, TC is the temperature coefficient and TA is the ambient temperature. A capacitive humidity sensor produces an output voltage proportional to the input voltage and the ambient relative humidity as shows by Eq. (8) that is based on the capacitance absolute permittivity and dielectric constant (Fraden, 2004).

V0 α Vi × %RH

(8)

where, V0 is the sensor output voltage, Vi is the sensor input voltage and RHis the relative humidity. For the wireless networking, the short range communication uses ZigBee which is based on the IEEE-802.15.4 protocol with extended range and power. For best transmission the Fresnel Zone between the sender and receiver is estimated as follows (Hebel & Bricker, 2010).



Fig. 2. Functional flow chart of the proposed system.

rm = 17.32 ×

dkm 4 × fGhz

(9)

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

81

Fig. 3. Decision algorithm for the operative situation perception, cover moving direction calculation, overloaded status estimation and further requered sensors activation processes.

where, ris the radius, d is the distance and f is the transmission frequency. The data collected by a sensor node is transmitted to a coordinator through ZigBee. Using the long range communication GPRS the data is transmitted to the server from the coordinator. GPRS is a data network that overlays a GSM network. This data overlay network provides packet data transport at a higher rate and multiple users can share the same air-interface resources simultaneously. In the server the data is stored, processed and represented to serve the monitoring and optimization purpose. This study emphasizes on real time collection of sensors data and short range and long range communications to transmit these data. 3.2. Physical architecture The physical architecture is designed in three main sections from a bin to the control station such as smart bin, gateway and control station as shown in Fig. 1. The jobs are to acquire various bin condition data in real time, to transmit the data to the control station via gateway and to represent the data in a user friendly manner to monitor the bin status. The system is based on web-access architecture of a network for distributed bins. To accomplish the jobs, the core activities of the system are to collect sensor data instantly after wastes are thrown inside the smart bin, create and transmit data frame to the gateway from the bin through ZigBee, transmit the data from gateway to control station through GPRS, and store and process the data to represent the updated status of the bin. The smart bin is composed of sensor node with a set of sensory elements for real time data collection and transmission. For efficient functioning, six different sensors have been used to design a smart bin such as accelerometer, Hall effect, ultrasound, load cell, temperature and humidity sensors. The accelerometer sensor reacts as soon as the bin cover is opened. The Hall effect sensor keep tracks whether a bin cover is open or close. The other sensors start operation when the bin cover is opened and then closed. The ultrasound sensor measures the filling level inside the bin. The load cell sensor measures the weight of the waste inside the bin. The temperature and

humidity sensors measures the temperature and humidity respectively to inform the ambient condition of the bin owing to ensure about the proper functioning of other sensors. For the short range data communication, a ZigBee transceiver is used in the smart bin. The gateway receive the data send by the ZigBee transceiver and stores the data to its local database as well as sends the data to the control station through GPRS communication using a GSM/GPRS module. The control station contains the central server which hosts the database and DBMS. It receives the data sent by the gateway and stores it to the database. These data are then used further by the control station to route optimization program for routing optimization of waste collection vehicle. The central server also hosts web based user interface for bin status monitoring, updated route presentation and user interaction with the system. In this way, the status of a bin can be updated on real time with high precision. 3.3. Operational principle At idle time, the sensor node in a smart bin remains in sleeping state except the accelerometer sensor. The functioning of the system starts when someone opens the bin cover to throw waste. To clarify the operational principle of the system, it’s functional flowchart and intellegent decision algorithm is defined in the following two sections. 3.3.1. Functional flowchart Fig. 2 shows the functional flowchart of the designed system. As soon as someone opens the bin cover or picks up the bin for unloading, the accelerometer sensor sense the values of acceleration in all three directions and percieves the operative situation. There are three operative situations such as loading, unloding and no operation. In case of the first two, the accelerometer interrupts the mote to wake up. After awaken, the mote activates the hall effect sensor to calcutate the cover moving status as well as it’s overloaded status. Based on whether the bin is overloaded or not, the mote activates others

82

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

Fig. 4. (a) 3 complete Experimental Model (EM) and (b) Placement and installation of sensory elements in EM.

different numbers of sensors for different perception. Then, the activated sensors acquire corresponding values related to bin status parameters. After collecting all the data, the mote creates frame comprising the sensors data and the node information and send it to the ZigBee transchiver module. The module creates ZigBee frames and send it to the gateway. The gateway which is acting as the sensor network coordinator node, receives the ZigBee frames and stores the data into it’s local database. At the same time, it opens transmission channel using TCP/IP through the GPRS and sends request to connect with the server. A multi-threaded background process called daemon tools is always running on the server which receives the connections

requests and listens for incoming data from the gateways. At the control station, the server receives and stores the data to a central database and updates specific bin status. 3.1.2. Rule based decision algorithms For the sub-processes of intelligent decisive process of Fig. 2, a set of rule based algorithm has been developed to percieve and decide the operative situation, bin cover moving direction, overloaded status and further required sensors activation. Fig. 3 shows a flow diagram of the decision process for the four sub-processes of Fig. 2. According to Fig. 3, after getting an interrupt by the acceleratometer, the mote wakes up and takes some decisions about the bin

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

83

Fig. 5. Control station architecture.

conditions before acquiring the bin parameters by using the following rule based algorithms. Here, Ax , Ay and Az are the changes of acceleration due to the changes of magnitude of the x, y and z axis of the accelerometer, Vh is the Hall voltage, tw is the waiting time and tth is the time threshold. So, the rule based decision algorithms for the above mentioned four sub-processes are as follows.

Rule 1: if (Ax , Ay , Az and Vh = 0 or 1) then operative situation = unloading Rule 2: if (Ay , Az and Vh = 1) then operative situation = loading AND cover moving direction = opened from closed position Rule 3: if (Ay , Az and Vh = 0) then operative situation = loading AND cover moving direction = closed from opened position Rule 4: if (Rule 2 = = TRUE and tw > tth ) then bin condition = overloaded Rule 5: if (Rule 2 = = TRUE) then If (Rule 3 = = TRUE and tw < = tth ) then bin condition = not overloaded Rule 6: if (Rule 4 = = TRUE) then actives sensors = load cell, temperature, humidity Rule 7: if (Rule 5 = = TRUE) then actives sensors = ultrasonic, load cell, temperature, humidity

4. Devices and experimental setup The real time bin monitoring system is established on a web based platform for WSN. Three core parts represent the system, such as (i) smart bin attached with a set of sensors and a ZigBee transceiver, (ii) gateway with local database, ZigBee transceiver and GSM/GPRS module, and (iii) control station interfaced with GSM/GPRS receiver, central data storage with database management system and a set of back-end applications. The selected devices, their calibration process and experimental setup are described in the following sections. 4.1. Smart bin For the experimental purpose, three fully equipped prototype of smart bin called experimental model (EM) have been designed as shown in Fig. 4(a). Customized version of 240 L two wheels standard bins is used that are most common within the experimented area. A set of cautiously selected sensors are used to furnish the bin. Among the sensors, some are installed on underside of the bin’s cover that contains accelerometer, temperature, humidity, Hall effect and ultrasonic sensors. The load cell sensor is mounted on a foil attached under the bin. To integrate the sensors, Smart Metering sensor board is used with Waspmote (Waspmote, 2014) that acts as a sensor node. The waspmote uses ATmega1281 chip as MCU and one ZigBee RF module is attached with it for data communication. The sensor node is power up by a rechargeable battery. The sensor node with all the sensors except the load cell is placed inside an 8 × 6 × 2.5 cm box mounted under the cover. The load cell is connected with the sensor board through a

84

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

Table 2 Measurement errors of ultrasonic sensor. Test no.

Measured filling level (cm)

Actual filling level (cm)

Error (%)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

84.33 84.15 83.77 82.22 81.03 78.41 76.8 75.21 70.3 69.3 67.72 67.33 61.81 55.2 55.08 54.67 53.81 52.06 52.05 49.12 49.05 43.2 41.48 39.21 38.27 36.45 35.45 34.76 26.86 21.59 19.49 15.78 12.35 11.92 7.57 5.55

80 80 79 77.5 77.5 74 73.5 72 67 66 64 64 58.5 52 52 50 58 54 48.5 46 46 39 39 37 35 35 32 32 23 20 18 14.15 11.2 10.9 6.9 5

5.13459 4.93167 5.694163 5.740696 4.356411 5.624283 4.296875 4.268049 4.694168 4.761905 5.493207 4.945789 5.355121 5.797101 5.591866 8.542162 −7.78666 −3.72647 6.820365 6.351792 6.218145 9.722222 5.978785 5.636317 8.544552 3.978052 9.732017 7.940161 14.37081 7.364521 7.644946 10.32953 9.311741 8.557047 8.850727 9.90991

wire attached behind the bin. All these integration and placement of sensing devices are shown in Fig. 4(b). Among the sensory element, LIS331DLH 3D motion sensor (LIS331DLH, 2009), is employed as an accelerometer. It has full scales of ± 2 g/±4 g/±8 g and consumes very low power while giving high performance. For the Hall effect sensor, PLA41201 is installed that uses U625000 as permanent magnet (PLA41201, n.d). It has a sensibility up to 13 mm and can produce a voltage between 0 V and 1 V depending on the distances between the conductor and magnet. The linear active thermistor integrated circuit MCP9700A is used as temperature sensor. It has a temperature measurement range of −40 °C to +125 °C with an accuracy of ± 2 °C while requires very low operating current of 6 μA. The 808H5V5 module is employed as humidity sensor which is designed based on capacitor polymer sensor. It can measure the ambient relative humidity with a range of 0–100%RH and has an accuracy of ±2%RH to ±4%RH. The ultrasonic sensor XLMaxSonar-WRA1 (Ultrasonic Sensor, n.d.) is selected as level measurement sensor that is weather resistant and can be used at outdoor environment. It has a resolution of 1 cm with a detection range of 765 cm and consumes low power. The resistance AMS Load Cell from Hanyu is used as load sensor (AMS load sensor, n.d.). It has a sensitivity of 2.0 ± 01 mv/V and accuracy grade is about 0.02% of full scale. It is weather resistant and fit for use outdoor environment. XBee-ZB-PRO (S2) RF transceiver is employed as radio module that conforms the ZigBee-PRO v2007standard (XBee-PRO ZB, 2014). It has an outdoor RF range up to 2 miles (3200 m) considering the line-of-sight. To supply power to the sensor node, a 2300 mAh Li-ion rechargeable battery is used for the EM. In case of the design and installment of different sensing devices for the EM of smart bin, details of the manufacturing issues have been

Fig. 6. (a) Smart bin positioned beside the Engineering faculty and (b) Gateway mounted beside the Sistem Pinter lab.

over looked. Preserving the basic structure and functionality, the EM contains all the devices but is not embedded for fulfilling the requirements to work in challenging environmental situations and not to be suitable for regular waste collection jobs, such as unloading by waste collection vehicles and operating by the workers. The EM is used for a preliminary feasibility evaluation, i.e. to estimate bin content by examine the signal quality of the sensory equipment. 4.2. Gateway Meshlium (2014) from libelium is used as the gateway. It can integrate different communication interfaces with having its own storage. It provides embedded solutions based on Linux OS that offers a commanding support setting to facilitate the development of costeffective wireless machine-to-machine applications that require different communication technologies. In this application, the gateway is used as a ZigBee to GPRS router for Waspmote nodes and thus, XBee-ZB-PRO (S2) RF transceiver module is used in the gateway to receive the ZigBee data from the sensor nodes and sim900 GSM/GPRS module from sim900 (2012) is used to provide the long range communication between the gateway and the control station. All the networking options can be controlled either from web interface or from SSH console. The data coming from the sensor nodes can be stored in the local database as well as can be exported to an external database that is connected to internet. 4.3. Control station The control station is the junction for the incoming data from all the gateways. The architecture of the control station is as shown in Fig. 5. The control station contains servers that establish connection with the gateways, receive and store data, analyze the data for route optimization and facilitate user friendly interfaces and graphical outcomes to the administrators and users of the system. A daemon process runs in background always to manage all the connection requests from the gateways. The open-source relational database management system MySQL is employed to store and manage the data. The web application is designed using HTML, PHP and JavaScript for representing the status of each bin and updated route based on the updated bin status information. 4.4. Calibration and pilot setup Before the experimental setup each bin is calibrated and offset errors are removed in order to confirm readings from a sensor are consistent, to estimate the accuracy of reading and to conclude the reliability of the sensor. For the accelerometer, it produces some

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

85

Fig. 7. Web based GUI of the developed system. Table 3 A sample snap of collected raw data from three EMs.

Bin ID Secret ID Bin_2 Bin_1 Bin_1 Bin_3 Bin_1 Bin_3 Bin_2 Bin_1 Bin_2 Bin_1

35689659

Type of Frame X axis Y axis Z axis acTimestamp frame number acceleration acceleration celeration

2014-02-07 12:02:23 35690575 2014-02-07 11:55:17 35690575 2014-02-07 11:40:02 35690098 2014-02-07 11:33:47 35690575 2014-02-07 11:23:32 35690098 2014-02-07 11:08:12 35689659 2014-02-07 11:07:22 35690575 2014-02-07 10:57:05 35689659 2014-02-07 10:43:20 35690575 2014-02-07 10:35:13

Hall effect sensor response

Waste filling Weight of level (cm) waste (kg)

Temperature (°C)

Humidity (%)

Available battery power (%)

253

3

−21

−211

−965

0

55.23

3.15

26.41

60.18

93

253

5

−337

−572

−1075

1

–

–

25.89

70.03

56

131

4

6

−293

−773

1

9999

5.42

28.47

74.09

56

253

2

−20

−361

−925

0

27.12

1.03

29.78

68.11

75

253

3

69

−344

−741

0

54.67

2.03

28.29

73.56

50

253

1

−12

−463

−855

0

0

0

29.22

69.19

75

253

2

4

−327

−894

0

31.59

1.78

25.12

61.05

92

253

2

78

−305

−749

0

38.27

1.06

28.31

72.98

49

253

1

-25

−311

−921

0

0

0

25.15

59.72

94

253

1

-45

−336

−540

0

0

0

28.19

73.61

53

offset errors in steady state rather than the standard output of (0,0,-g) in (x,y,z) axis, respectively. For each bin, this error is corrected individually and also the effect of gravity acceleration is taken into account. The output of Hall effect sensor is calibrated to 0 or 1 in response to the produce voltage of 0 V or greater than 0 V. As a result, this sensor will produce an output of 0 when the conductor and the magnet is in contact and 1 otherwise. To calibrate and validate the ultrasonic sensor, it was placed in the upside of an empty bin of depth 85 cm and moved towards the bin floor with a step of 1 cm. Though it produces an accurate result for the empty flat surface, but it gives some erroneous output for bins with solid waste due to the irregularities of surface. Table 2 shows a comparison of measured filling level

value and actual filling level value during the testing process. The assessment highlights that, the percentage of error reading is between 5% and 10% for more cases. It also indicates that, the trend of error is reduced as filling level is increased which allows the opportunity to use this sensor with an aim to fulfill the goal of the system. The load cell sensor is tested by using known weight of 500 g, 1 kg, 2 kg and 5 kg. For the test run, the experimental area selected was besides the old building of engineering faculty of University Kebangsaan Malaysia. One EM is positioned beside the canteen and another two EM is placed beside the street. The gateway is mounted on top beside the Sistem Pinter lab. The server is designed in the

86

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

100 80 60 40 20 0

Empty Reading_1 Reading_2 Reading_3 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Reading_9 Reading_10 Reading_11 Reading_12 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Unloading

Filling Level (cm)

a 120

Round_1

Round_2

Round_3

Round_4

Round_5

Round_6

Round

12 10 8 6 4 2 0

Empty Reading_1 Reading_2 Reading_3 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Reading_9 Reading_10 Reading_11 Reading_12 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Unloading

Weight (Kg)

b 14

Round_1

Round_2

Round_3

Round_4

Round_5

Round_6

32 31 30 29 28 27 26 25 24

Empty Reading_1 Reading_2 Reading_3 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Reading_9 Reading_10 Reading_11 Reading_12 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Unloading Empty Reading_1 Reading_2 Reading_3 Reading_4 Reading_5 Reading_6 Reading_7 Reading_8 Unloading

Temperature (0C)

c 33

Round_1

Round_2

Round_3

Round_4

Round_5

82 80 78 76 74 72 70 68

Relative Humidity (%)

Round

Round_6

Round Temperature

Humidity

Fig. 8. (a) Filling level readings of 6 complete rounds for Bin_1, (b) Weight measurement readings of 6 complete rounds for Bin_1 and (c) Temperature and relative humidity readings of 6 complete rounds for Bin_1.

lab. Fig. 6 (a) and (b) shows the pilot setup for the smart bin and gateway. 5. Experimental results and discussions To test the performance of the designed system, about 300 test runs have been performed using the three EMs. Students are invited to throw solid waste packed in bags. Several complete rounds have been executed where a full round comprises of fill up an empty bin with solid waste by a number of waste throwing operations and then unload the overloaded bin. During the test runs, the data of each sensor for every bin have been recorded and stored in the database. A sample snap of the table that contains the collected data is shown in Table 3.

The developed web application provides the real time bin status and presents the updated bin filling level by using a bar graph in every 5 s. Fig. 7 shows the GUI of the bin monitoring system web application. The users can monitor the bins status in various combinations by selecting different options. They can monitor the status of a specific bin or all bins in together. The users can also view a specific sensor or combination of sensors, or all sensors for each bin combination. Using any one of the above grouping, the users can observe the status by both in raw data format or by graphical format. Fig. 8 shows the test run results for Bin_1. During the experiment, 6 complete rounds have been completed for Bin_1. The figures show the corresponding sensors outputs for each waste throwing operation that is called by reading. The sensors outputs are plotted for each

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88 Table 4 Communication performance analysis from bin to gateway. Attribute

Bin_1

Bin_2

Bin_3

Transmit power level Distance to gateway Line of sight No. of transmitted packets Average RSSI to gateway Throughput

10 dBm 85 m No 100 −73 dBm 100%

10 dBm 130 m No 100 −79 dBm 100%

10 dBm 170 m No 100 −83 dBm 93%

reading. For the filling level status, the result shows the bin fill level after each waste throwing operation inside the bin. The bin’s depth is 85 cm and 100 cm in the chart represent overloaded status. For the test runs, the unloading operation is performed when it is almost full or overloaded. The load cell output of each reading is increased with each waste throwing operation as expected and the amount is depended on the type of material of the waste. The readings for temperature and humidity sensors are significant as they represent the usual environmental condition for correct operation of the others sensing devices. By observing the readings for these two sensors, the ambient condition of the smart bin can be perceived. The experimented area was surrounded by various obstacles like trees, concrete walls, and wooden doors between the bins and the gateway. About 300 test runs have been performed considering 100 runs for each bin. The experimented results show that XBee-Pro S2 ZigBee modules are consistent due to the existence of obstacles in outdoor environment. The recorded results presents excellent throughput for a transmit power level of 10 dBm in various distances. The summary of the communication performance from bin to gateway is shown in Table 4.

6. Conclusion Solid waste management suffers from the lack of throughput, shortage of required data regarding solid waste, efficiency problems, collection delays and resistance to new technologies. One commonly cited means to overcome this problem is through advanced information and communication technologies. However, the challenge arises when a conclusion has to be made to select the best technology for the fulfillment of present needs because each technology has its own technical, economic, and risk considerations. At present waste management is a major concern for the authorities responsible for this task. Waste collection is a costly service that consumes most of the municipality budget. It also creates a huge environmental impact. This paper introduced a real time bin monitoring system framework by using various sensing and communication technologies integrated with intelligent decision algorithms. To achieve the proposed system, theoretical models, architectural layout and rule based decision algorithm have been developed. The experimental results show that the developed system responded as soon as someone throws waste inside a smart bin and transmits the updated bin status to the control station via the sensor network coordinator named gateway. The control station used the real time bin information to show their status as well as can feed the data to an optimization model for route optimization. The main contribution of this study in the field of expert and intelligent system is the design and development of an automated bin status monitoring system that exercises a set of novel rule based decision algorithms. The algorithms facilitate the data acquisition process in the smart bin real time and fully automated. Thus, the system provides a robust platform that can acquire bin condition data instantly as soon as waste is thrown inside a smart bin. As a result, the developed system successfully overcomes the problems of existing systems. It offers an appropriate estimation of bin status while poses

87

less operation costs. It also ensures the best utilization of power thus making the system energy efficient. However, some limitations in the present prototypes are identified, especially regarding the ultrasonic sensor that sometimes produces erroneous output due to the irregularities of solid waste patterns. Also, the load cell sensor occasionally gives wrong output for light waste material due to the installation of the load cell sensor with the foil. Another limitation is that, although the practical solid waste management services utilize different kinds and sizes of trash bins, this study only employed 240 L two-wheeled bin because of the time constraints and the inability to study and to investigate other types and sizes. This study presented a general model for real time bin status monitoring. As such, there are many areas that can be improved for successful real world waste collection. Some of the recommendations and suggested areas for future research are: firstly, GPS and GIS should be integrated with the system to make it more expert for practical waste collection operation. Secondly, waste collection field is generally wide, different wastes has different bins which are practically impossible to consider waste collection under one bin type. So, the hardware integration and decision algorithms should be revised for different sizes and types of waste bins to make the system more intelligent and expert. Thirdly, the system should be tested in a small scale municipality. The municipality could install some smart bins and identifies real advantages and constraints of the system. The municipality could as well benchmark and examine the performance and results of the bin status monitoring system in order to truly verify the feasibility of such implementation. Finally, the possibility of others long range communication technologies such as 3G or LTE in the gateway should be experimented. The system aims to provide real time bin status information to the city administrations to serve the citizens with an efficient and effective manner. Still, the design focuses mostly on integrating different sensing and communication technologies with intelligent decision algorithms. Development of applications for the citizen, recycling factories and other stakeholders is also a requirement for better service. We have evaluated the proposed system and shown that developing a real time monitoring system can help to implement dynamic routing for waste collection which in turn can give a significant increase of cost effectiveness, which is one of the most indicating criteria for a Smart City. Acknowledgment The authors acknowledge the financial support from Ministry of Higher Education, Malaysia, Grants LRGS/TD/2011/UKM/ICT/04/01 and PRGS/1/12/TK02/UKM/02/2. References Alagöz, A. Z., & Kocasoy, G. (2008). Improvement and modification of the routing sys˙ tem for the health-care waste collection and transportation in Istanbul. Waste Management, 28(8), 1461–1471. AMS load sensor (n.d.). AMS load sensor specifications. . Accessed 14.05.15. Andrejašic, M. (2008). Mems accelerometers. University of Ljubljana. Faculty for mathematics and physics, Department of physics, Seminar. Arebey, M., Hannan, M. A., Basri, H., Begum, R. A., & Abdullah, H. (2011). Integrated technologies for solid waste bin monitoring system. Environmental Monitoring and Assessment, 177(1-4), 399–408. Arebey, M., Hannan, M., & Basri, H. (2013). Integrated communication for truck monitoring in solid waste collection systems. Advances in visual informatics (pp. 70–80). Springer. Arebey, M., Hannan, M. A., Begum, R., & Basri, H. (2012). Solid waste bin level detection using gray level co-occurrence matrix feature extraction approach. Journal of Environmental Management, 104, 9–18. Aravossis, K., Nikolaidou, E., & Fountzoula, C. (2015). Solid waste management through a modern innovative PAYT system. . Accessed 15.11.15.

88

Md.A.A. Mamun et al. / Expert Systems With Applications 48 (2016) 76–88

Asimakopoulos, G., Christodoulou, S., Gizas, A., Triantafillou, V., Tzimas, G., Viennas, E., Gialelis, J., Karadimas, D., & Papalambrou, A. (2015). Towards a dynamic waste collection management system using real-time and forecasted data. In Proceedings of the 16th international conference on engineering applications of neural networks (INNS) (p. 36). Badran, M. F., & El-Haggar, S. M. (2006). Optimization of municipal solid waste management in Port Said – Egypt. Waste Management, 26(5), 534–545. Catania, V., & Ventura, D. (2014). An approach for monitoring and smart planning of urban solid waste management using smart-M3 platform. In Proceedings of 15th conference of open innovations association FRUCT (pp. 24–31). Chang, N.-B., & Wei, Y. L. (1999). Strategic planning of recycling drop-off stations and collection network by multiobjective programming. Environmental Management, 24(2), 247–263. Chen, T.-C., & Lin, C.-F. (2008). Greenhouse gases emissions from waste management practices using life cycle inventory model. Journal of Hazardous Materials, 155(1–2), 23–31. Chen, T., & Li, X. (2010). Municipal solid waste recycle management information platform based on internet of things technology. In Proceedings of international conference on multimedia information networking and security (MINES’10) (pp. 729– 732). Cheng, S., Chan, C. W., & Huang, G. H. (2003). An integrated multi-criteria decision analysis and inexact mixed integer linear programming approach for solid waste management. Engineering Applications of Artificial Intelligence, 16(5), 543–554. Chowdhury, B., & Chowdhury, M. U. (2007). RFID-based real-time smart waste management system. In Proceedings of IEEE Australasian telecommunication networks and applications conference (ATNAC 2007) (pp. 175–180). Faccio, M., Persona, A., & Zanin, G. (2011). Waste collection multi objective model with real time traceability data. Waste Management, 31(12), 2391–2405. Fontaras, G., Martini, G., Manfredi, U., Marotta, A., Krasenbrink, A., Maffioletti, F., Terenghi, R., & Colombo, M. (2012). Assessment of on-road emissions of four Euro V diesel and CNG waste collection trucks for supporting air-quality improvement initiatives in the city of Milan. Science of The Total Environment, 426, 65–72. http://www.sciencedirect.com/science/article/pii/S0048969712003993. Fraden, J. (2004). Handbook of modern sensors: physics, designs, and applications. Springer. Fraile, J. A., Bajo, J., Corchado, J. M., & Abraham, A. (2010). Applying wearable solutions in dependent environments. IEEE Transactions on Information Technology in Biomedicine, 14(6), 1459–1467. Friedlos, D. (2005). Australian waste-collection businesses tag trash bins. RFID Journal (Abruf: 26.10. 2009), http://www.rfidjournal.com/article/view/5145, Accessed 14.05.23. Han, D.-M., & Lim, J.-H. (2010). Design and implementation of smart home energy management systems based on zigbee. IEEE Transactions on Consumer Electronics, 56(3), 1417–1425. Hannan, M., Arebey, M., Begum, R. A., & Basri, H. (2011). Radio Frequency Identification (RFID) and communication technologies for solid waste bin and truck monitoring system. Waste Management, 31(12), 2406–2413. Hannan, M., Arebey, M., Begum, R. A., & Basri, H. (2012). An automated solid waste bin level detection system using a gray level aura matrix. Waste Management, 32(12), 2229–2238. Hebel, M., & Bricker, G. (2010). Getting started with XBee RF modules. . Accessed: 24/04/2014. Huddleston, C. (2006). Intelligent sensor design using the microchip dsPIC. Newnes. Islam, M., Hannan, M., Basri, H., Hussain, A., & Arebey, M. (2014). Solid waste bin detection and classification using Dynamic Time Warping and MLP classifier. Waste Management, 34(2), 281–290. Isoaho, S., & Peltoniemi, M. (2004). Telematics for waste collection and transport. In Proceedings of 2nd international conference on ecological protection of the planet earth (pp. 109–115). Johansson, O. M. (2006). The effect of dynamic scheduling and routing in a solid waste management system. Waste Management, 26(8), 875–885. Khan, A. A. (1985). An improved linear temperature/voltage converter using thermistor in logarithmic network. IEEE Transactions on Instrumentation and Measurement, 4(5), 635–638. Khedo, K., Perseedoss, R., & Mungur, A. (2010). A wireless sensor network air pollution monitoring system. International Journals of Wireless and Mobile Networks, 2(2), 31– 45. Kietzmann, J. H. (2008). Interactive innovation of technology for mobile work. European Journal of Information Systems, 17(3), 305–320. Kistler, R., Bieri, M., Wettstein, R., & Klapproth, A. (2009). Tunneling smart energy protocols over ZigBee. In Proceedings of IEEE conference on emerging technologies & factory automation (ETFA 2009) (pp. 1–8). Kulcar, T. (1996). Optimizing solid waste collection in Brussels. European Journal of Operational Research, 90(1), 71–77. Lee, J. A., & Thomas, V. M. (2004). GPS and radio tracking of end-of-life products [recycling and waste disposal applications]. In Proceedings of IEEE international symposium on electronics and the environment (pp. 309–312). Lei, M., Chuanhua, L., Yuezhao, Z., Haijun, C., & Yanghuiqin, D. (2011). An environment friendly energy recovery technology: municipal solid waste gasification. In Proceedings of third international conference on measuring technology and mechatronics automation (ICMTMA’11) (pp. 407–409). Lin, H.-Y., Chen, G.-H., Lee, P.-H., & Lin, C.-H. (2010). An interactive optimization system for the location of supplementary recycling depots. Resources, Conservation and Recycling, 54(10), 615–622.

LIS331DLH (2009). MEMS digital output motion sensor ultra low power high performance 3-axes nano accelerometer. . Accessed 14.05.15. Longhi, S., Marzioni, D., Alidori, E., Di Buo, G., Prist, M., Grisostomi, M., & Pirro, M. (2012). Solid waste management architecture using wireless sensor network Technology. In Proceedings of 5th international conference on new technologies, mobility and security (NTMS’12) (pp. 1–5). Lu, J.-W., Chang, N.-B., & Liao, L. (2013). Environmental informatics for solid and hazardous waste management: advances, challenges, and perspectives. Critical Reviews in Environmental Science and Technology, 43(15), 1557–1656. Lyshevski, S. E. (2002). MEMS and NEMS: systems, devices, and structures. CRC Press. Mainwaring, A., Culler, D., Polastre, J., Szewczyk, R., & Anderson, J. (2002). Wireless sensor networks for habitat monitoring. In Proceedings of the 1st ACM international workshop on wireless sensor networks and applications (pp. 88–97). McLeod, F., Erdogan, G., Cherrett, T., Bektas, T., Davies, N., Speed, C., Dickinson, J., & Norgate, S. (2013). Dynamic collection scheduling using remote asset monitoring: case study in the UK charity sector. Transportation Research Record: Journal of the Transportation Research Board, 2378(1), 65–72. McLeod, F., Erdogan, G., Cherrett, T., Bektas, T., Davies, N., Shingleton, D., Speed, C., Dickinson, J., & Norgate, S. (2014). Improving collection efficiency through remote monitoring of charity assets. Waste Management, 34(2), 273–280. Meshlium (2014). Meshlium xtreme – datasheet v4.3. . Accessed 14.05.10. Municipal Solid Waste (2012). U.S. environmental protection agency. . Accessed 14.08.04. Muthukumaran, P., & Sarkar, S. B. (2013). Solid waste disposal and water distribution system using the mobile adhoc network. In Proceedings of the 2013 international conference on emerging trends in communication, control, signal processing & computing applications (C2SPCA) (pp. 1–4). Nielsen, I., Ming, L., & Nielsen, P. (2010). Optimizing supply chain waste management through the use of RFID technology. In Proceedings of IEEE international conference on rfid-technology and applications (RFID-TA’10) (pp. 296–301). Noche, B., Rhoma, F. A., Chinakupt, T., & Jawale, M. (2010). Optimization model for solid waste management system network design case study. In Proceedings of the 2nd international conference on computer and automation engineering (ICCAE’10) (pp. 230– 236). O’Connor, M. (2007). Greek RFID pilot collects garbage. RFID Journal, http://www. rfidjournal.com/article/view/2973, Accessed 14.05.22. O’Connor, M. (2008). Routeware launches RFID solution for waste haulers. RFID Journal http://www.rfidjournal.com/articles/view/4067/, Accessed 14.05.22. Papalambrou, A., Karadimas, D., Gialelis, J., & Voyiatzis, A. (2015). A versatile scalable smart waste-bin system based on resource-limited embedded devices. In Proceedings of the 2015 IEEE 20th conference on emerging technologies & factory automation (ETFA) (pp. 1–8). Parvis, M., & Carullo, A. (2001). An ultrasonic sensor for distance measurement in automotive applications. IEEE Sensors Journal, 1(2), 143–147. PLA41201 (n.d). Magnetic proximity sensor: Wce-electronics product ref no. 17244. . Accessed: 14/06/2014. Ramsden, E. (2011). Hall-effect sensors: theory and application. Newnes. Reis, P., Caetano, F., Pitarma, R., & Gonçalves, C. (2015). iEcoSys–an intelligent waste management system. In Proceedings of new contributions in information systems and technologies (pp. 843–853). Springer. Reis, P., Pitarma, R., Goncalves, C., & Caetano, F. (2014). Intelligent system for valorizing solid urban waste. In Proceedings of the 2014 9th Iberian conference on information systems and technologies (CISTI) (pp. 1–4). Rovetta, A., Xiumin, F., Vicentini, F., Minghua, Z., Giusti, A., & Qichang, H. (2009). Early detection and evaluation of waste through sensorized containers for a collection monitoring application. Waste Management, 29(12), 2939–2949. sim900 (2012). SIM900_Specification_V1203. . Accessed 14.07.05. Swedberg, C. (2009). Mississippi county tracks waste pickup. RFID Journal Web:, http: //www.rfidjournal.com/articles/view/5263, Accessed 14.05.26. Ultrasonic Sensor (n.d.). XL-MaxSonar – WR/WRC series, high resolution, IP67 weather resistant, ultra sonic range finder. . Accessed 14.05.14. Vacuum system history (n.d.). Centralsug: the first vacuum system ever. . Accessed14.04.17. Vicentini, F., Giusti, A., Rovetta, A., Fan, X., He, Q., Zhu, M., & Liu, B. (2009). Sensorized waste collection container for content estimation and collection optimization. Waste Management, 29(5), 1467–1472. Wang, N., Zhang, N., & Wang, M. (2006). Wireless sensors in agriculture and food industry—recent development and future perspective. Computers and Electronics in Agriculture, 50(1), 1–14. Waspmote (2014). Waspmote – datasheet v5.0. . Accessed 14.04.15. Wilson, B. G., & Vincent, J. K. (2008). Estimating waste transfer station delays using GPS. Waste Management, 28(10), 1742–1750. World Bank (1999). What a waste – solid waste management in Asia: urban and local government working papers. . Accessed 14.03.03. XBee-PRO ZB (2014). Digi international inc.: XBee/XBee-PRO ZB RF modules. . Accessed 14.08.12. Yu, L., Wang, N., & Meng, X. (2005). Real-time forest fire detection with wireless sensor networks. In Proceedings of international conference on wireless communications, networking and mobile computing (pp. 1214–1217).