Top–k Query based Dynamic Scheduling for IoT ...

Top–k Query based Dynamic Scheduling for IoT-enabled Smart City Waste Collection Theodoros Anagnostopoulos1, Arkady Zaslavsky2,1, Alexey Medvedev1, Sergei Khoruzhnicov1 1

Department of Infocommunication Technologies, ITMO University, St. Petersburg, Russia [email protected], [email protected], [email protected] 2 CSIRO Computational Informatics, CSIRO, Box 312, Clayton South, Vic, 3169, Australia [email protected]

Abstract – Smart Cities are being designed and built for comfortable human habitation. Among services that Smart Cities will offer is the environmentally-friendly waste/garbage collection and processing. In this paper, we motivate and propose an Internet of Things (IoT) -enabled system architecture to achieve dynamic waste collection and delivery to processing plants or special garbage tips. In the past, waste collection was treated in a rather static manner using classical operations research approach. As proposed in this paper, nowadays, with the proliferation of sensors and actuators, as well as reliable and ubiquitous mobile communications, the Internet of Things (IoT) enables dynamic solutions aimed at optimizing the garbage truck fleet size, collection routes and prioritized waste pick-up. We propose a top−𝒌 query based dynamic scheduling model to address the challenges of near real-time scheduling driven by sensor data streams. An Android app along with a user-friendly GUI is developed and presented in order to prove feasibility and evaluate a waste collection scenario using experimental data. Finally, the proposed models are evaluated on synthetic and real data from the city municipality of St. Petersburg, Russia. The models demonstrate consistency and correctness.

this paper does not intend to present all of them. It is used the definition best suits to the IoT-enabled waste collection in smart cities, which is [10]: “A Smart City is a city well performing in a forward-looking way in the following fundamental components (i.e., Smart Economy, Smart Mobility, Smart Environment, Smart People, Smart Living, and Smart Governance), built on the ‘smart’ combination of endowments and activities of self-decisive, independent and aware citizens”. This definition contains the fundamental component of Smart Environment which is relevant to environmental pollution [11]. A municipality service which acts as a countermeasure to environmental pollution within the Smart City is the dynamic waste collection.

Keywords – Top−𝒌 Query; Dynamic Scheduling; IoT; Waste Collection; Smart City.

I. INTRODUCTION Smart Cities is the future of civil habitation, since by 2050 the vast amount of earth population (i.e., 70 percent) will move to urban areas thus forming vast cities [1]. These cities will incorporate smart infrastructure in order to manage their needs for fundamental and advanced services [2]. The use of Future Internet with the enhancement of IPv6 (i.e., 6LoWPN) as well as sensors and wireless sensor networks enable IoT to reform Smart City municipality activity in every aspect of daily life [3], [4]. One such a service that has great impact on citizen quality of life is the efficient waste collection [5]. In past years waste collection was treated in a rather static approach, nowadays with the proliferation of sensors and actuators; IoT enable dynamic solutions as well [6]. In order to understand in depth what the concept of a Smart City is, a definition must be provided. A lot of definitions have been provided in the literature [7], [8], [9],

Fig. 1. System Overview

In this paper we introduce an IoT-enabled system architecture to achieve efficient dynamic waste collection. We also propose a top−𝑘 query based dynamic scheduling model to face the demanding nature of scheduling timing [12], [13]. Finally, an Android app along with a user-friendly GUI is presented in order to evaluate a waste collection scenario on synthetic and real experimental data. This paper is structured as follows. Section II reports related work. Section III presents the IoT-enabled system architecture. Section IV describes the top−𝑘 query dynamic scheduling model. Section V presents evaluation performance

with the Android app, while Section VI concludes the paper and discusses future work. II. RELATED WORK We report on methods which adopt dynamic models for waste collection. In [14] authors introduce a dynamic routing model based on fuzzy demands by assuming the demands of the customers as fuzzy variables. Model incorporates a heuristic approach based on fuzzy credibility theory. In [15], authors propose routing with time windows which analyze the logistics activity within a city. Model finds the cost optimal routes in order the trucks to empty the bins with an adaptive large neighborhood search algorithm. Authors in [16] introduce a rollon-rolloff routing, serving multiple disposal facilities, with huge amounts of waste at construction sites and shopping districts. It is applied large neighborhood search with iterative heuristics algorithms. In [17] authors incorporate discrete event simulation for waste collection from underground bins. Model applies dynamic planning to exploit information transmitted through motion sensors embedded in the underground bins. In [18] authors propose a genetic algorithm to solve dynamic routing problem. Specifically, model assumes that the waste collection problem could be treated as a Traveling Salesman Problem (TSP). Then the genetic algorithm solves the TSP optimally. Authors in [19] propose a heuristic method for dynamic routing considering several tunable parameters. Sensors enable reverse inventory routing in more dense waste networks. Heuristics deal with uncertainty of daily and seasonal effects. Authors in [20] propose a routing model which incorporates Ant Colony System (ACS) algorithm in order to achieve dynamic routing. They treat the location of the bins as a spatial network and apply 𝑘 −means in order to cluster the bins distribution into a set of partial clusters. In [21] authors combine routing and scheduling optimization. Historical data applied to bins individually establish the daily circuits of collection points to be visited. Planning is applied to scheduling for better system management. Authors in [22] develop routing with a mobile measuring system on the trucks. They perform stochastic dynamic routing which makes corrections during or after the execution of the existing routes. Authors in [23] introduce a memetic algorithm to perform routing enforced with time windows and conflicts context. Model incorporates a combination of flow and set partitioning formulation to achieve multiobjective optimization. Authors in [24] consider dynamic scheduling over a set of previously defined collection trips. The main objective of the approach is to minimize the total operational and fixed truck costs. A mathematical formulation methodology is proposed in [25] developing a plan of service areas, defining routing, and designing scheduling taking into consideration possible new alternative solutions in managing the system as a whole. In [26] authors evaluate dynamic planning methods applied for waste collection of underground bins. Model reduces the amounts of carbon dioxide released in the environment from trucks by making dynamic routing more effective.

Model in [27] is specialized in waste collection of plastic waste which is differentiated from the other municipal solid waste. It is achieved heuristic redesign of the collection routes using an eco-efficiency metric balancing the trade-off between the costs and environmental issues. A heuristic solution is proposed in [28] where authors state the waste collection as a periodic truck routing problem with intermediate waste depots. The model incorporates variable neighborhood search and dynamic programming in order to achieve optimal solution. Dynamic scheduling and routing model in [29] applies capacity sensors and wireless communication infrastructure thus to be aware of the bins state. It incorporates analytical modeling and discrete-event simulation in order to achieve real-time dynamic routing and scheduling. Authors in [30] use improved dynamic route planning. They enhance a guided variable neighborhood threshold meta heuristic adapted to the problem of waste collection. In [31] authors propose a novel IoT-enabled dynamic routing model for waste collection in a Smart City. The proposed model is robust in case of emergency (i.e., a road under construction, unexpected traffic congestion). Finally, authors in [32] propose a robust waste collection model exploiting cost efficiency of IoT potentiality in Smart Cities. The research extends [31] by introducing a dynamic routing algorithm which is robust and copes with cases of truck replacement due to overload or damage in the city of St. Petersburg in Russia.

Fig. 2. System Architecture

Related research in waste collection focuses on dynamic scheduling and routing models. However less research states the waste collection as a Smart City service. Specifically, only in [31] and [32] it is addressed the waste collection as a problem which can be solved with IoT infrastructure; incorporated in Smart Cities. In this paper we extend the research proposed in [32] by introducing a system architecture for IoT-enabled waste collection in a Smart City. We also introduce a dynamic scheduling model which enhances top−𝑘 query and aggregates data generated by

sensors and actuators embedded on bins. The model is evaluated with an Android app along with a user-friendly GUI in order to interact and give routing directions to truck drivers while simultaneously receives feedback on real time.

engine which is implemented in OpenIoT. Data are stored in a spatial database in which mobile top−𝑘 queries specify the number of the full bins in order to initiate dynamic scheduling.

III. SYSTEM ARCHITECTURE

On certain timing; depending on dynamic scheduling, a dynamic routing is initiated, as described in [32], which incorporates GPS data embedded in trucks in order to provide drivers instructions and real time routing directions. Specifically, context aware information is available to drivers through an Android app along with a user-friendly GUI. The mobile application provides, in a Smart City Google Map, information about the top−𝑘 bins to be emptied as well as the routing trip should be traversed from the a truck in order to reach them. In Figure 2 it is presented the system architecture.

Assume a Smart City which incorporates IoT infrastructure for achieving efficient dynamic waste collection. Regarding spatial information the Smart City is divided into multiple sectors which cover the entire city area. Each sector contains a number of multiple intermediate waste depots, which are temporary waste storage areas. Out of the frontiers of the city there is located a number of multiple garbage tips used to store the waste collected from the depots. Further processing of the waste is performed by processing plants which are located near the garbage tips. The proposed system architecture incorporates a heterogeneous fleet of trucks for serving the waste collection infrastructure. Specifically, a fleet of Low Capacity Garbage Trucks (LCGTs) is used to collect waste from the bins located in the backyards of the sectors and store it temporarily to depots. A fleet of High Capacity Garbage Trucks (HCGTs) is used to collect waste from the depots and transfer it to the garbage tips. In this paper we are studying the special case of dynamic scheduling of waste from bins to depots through LCGTs which for reasons of simplicity would be stated as trucks. In Figure 1 it is presented the system overview.

Fig. 3. The relation 𝐵𝑖𝑛𝑠 with sample data

Fig. 4. Top−𝑘 query SQL statement

IV. TOP−𝑘 QUERY DYNAMIC SCHEDULING A. Relational Data Base Top−𝑘 queries are exploiting data stored in a relational database each time a truck has available capacity to load more waste from the bins. Let us assume that a bin has capacity 𝑐 while a truck has capacity 𝐶, then the truck can load waste form maximum 𝜏 = 𝐶/𝑐 bins. However in certain scheduling time for full bins 𝑘 it holds that 𝑘 ≤ 𝜏. In order to load the 𝑘 most full bins, in descending order, we apply top−𝑘 queries to the relational database. The structure of the database is composed of a single relation called 𝐵𝑖𝑛𝑠 which contains information about the GPS latitude and longitude of the bins and their capacity as it is updated online from the capacity sensors trough the IoT infrastructure. The 𝑖𝑛𝑑𝑒𝑥 of the relation is defined to be an ascending valued integer, while the rest of the data are defined to be real numbers. The database is simple and contains only one relation (i.e., 𝐵𝑖𝑛𝑠) because of mobile space limitations. Figure 3 depicts the relation 𝐵𝑖𝑛𝑠 with a sample set of data.

In the low level the system architecture is composed of a number of bins which are enabled with:    

RFIDs for identification tagging with 6LoWPAN, Capacity sensors for measuring the waste volume per bin, Actuators which lock the lids if a capacity threshold is reached, Wireless antennas to transmit sensor data to the system infrastructure.

All these data form a data fusion which has to be handled from the system in order to provide an advanced waste collection service. Cloud middleware is responsible to collect data from sensors, aggregate and clean them (i.e., discard or infer missing values) in order to provide them to the inference

Fig. 5. Applying top−𝑘 query for 𝑘 = 5

B. Top−𝑘 Query for Capacity Ranking In order to understand in depth the notion of top−𝑘 queries it could be assumed a relation as proposed (i.e., 𝐵𝑖𝑛𝑠). Then for a certain scheduling time the capacity of the

relation is updated on real time after new waste enter the bin through the mobile IoT infrastructure. A top−𝑘 query is an advanced SQL selection query which is enhanced with a descending ranking order by statement on the capacity data values. It is selected the top−𝑘 bins of the relation. Value of 𝑘 is application specific and depends on the capacity of the bin 𝑐, the capacity of the truck 𝐶 as well as scheduling time limitations (i.e., empty waste during a certain time threshold in order to avoid pollution effects). Figure 4 depicts the SQL statement of a top−𝑘 query. TABLE I. DYNAMIC SCHEDULING ALGORITHM

V. EVALUATION PERFORMANCE We evaluate dynamic scheduling with synthetic and real data in order to assess its performance. It is assumed that the Smart City is divided to 10 sectors. Each sector contains 50 bins. Totally the city has 500 bins. There are 2 depots attached to every sector. Each sector has 3 trucks to collect waste from the 𝑘 red bins and empty them to the depots. It is assumed that each bin has capacity 𝑐 = 100 kg, while each truck capacity is set to 𝐶 = 5000 kg. The number 𝑘 of the red bins is application specific and is dynamically set to 𝑘 ∈ {20, 25, 30, 35, 40, 45, 50}.

Input: 𝐵𝑖𝑛𝑠 Output: 𝑘 While (𝑡𝑟𝑢𝑒) Do 𝑘 = top−𝑘 query from relation 𝐵𝑖𝑛𝑠 /*Select 𝑘 red bins to load*/ End While

Bins are further categorized to three classes according to their capacity levels. There is the low level which is depicted with green color and implies that the bin is almost empty. There is the medium level which is depicted with yellow color and implies that the bin is half full. Finally there is the high level which is depicted with red color and implies that the bin is full and needs to get empty immediately by the truck. Top−𝑘 query returns only the red bins w.r.t capacity ranking which mean that 𝑘 is set properly (i.e., by the application) to select the high leveled, full of waste, bins. Figure 5 presents the selected data after applying a top−𝑘 query to the relation 𝐵𝑖𝑛𝑠 for 𝑘 = 5. C. Dynamic Scheduling Algorithm The dynamic scheduling algorithm locates the first available truck which can load waste from the 𝑘 red bins. Then it is performed a top−𝑘 query which exploits real time data from the relation 𝐵𝑖𝑛𝑠. The value of 𝑘 is application specific. The output of the dynamic scheduling algorithm is the latitude and longitude of the 𝑘 bins that satisfy the top−𝑘 query criteria. The 𝑘 bins are consequently input to the dynamic routing algorithm, as described in [32], in order the truck to perform a route and empty the waste to the first available depot in the Smart City. Dynamic scheduling algorithm is presented in Table I.

Fig. 6. Dynamic route collection of 𝑘 red bins

Fig. 7. Scheduling waste collection time

An Android app is implemented in order the drivers to have a user friendly GUI interface with the IoT system. It is incorporated Google Maps from the city of St. Petersburg in Russia. The relational database is implemented in Microsoft Access while the code is written in Eclipse Java environment. Latitude and longitude GPS data of the bins location are real from the Google Maps while the data used for the weights of the model are synthetic. In Figure 6 it is observed a dynamic route, as described in [29], initiated to collect the 𝑘 bins from a sector of St. Petersburg. Note that since there are 10 bins in total; only 5 are marked as red, thus being full. It is assumed that in this example 𝑘 = 5. Dynamic routing is initiated only for these 5 red bins, while dynamic scheduling evaluates online the status of the rest 2 yellow and 3 green bins. We expand our model with higher values of 𝑘 ∈ {20, 25, 30, 35, 40, 45, 50} as it is also described previously. In order to prove the superiority of our proposed dynamic scheduling model we compare it with a static model available in the literature [19]. The proposed model is proved to be superior. Specifically, we experimented on dynamic vs. static scheduling; with regards to waste collection time, for different values of 𝑘 bins. The results are presented on average for all the sectors of the Smart City, see Figure 7. It can be observed that scheduling time is varied with respect to the value of top−𝑘 bins; thus the higher the 𝑘 value the more time, logarithmically, required initiating a scheduling. Note that the higher scheduling time is reached for the static scheduling; since in this case there is no information about

the capacity of the collected bin (i.e., if it is full or half full), thus leading to low system performance.

logarithmically, required computing a dynamic scheduling. Static scheduling seems to be more efficient with regards to CPU overhead cost; since it uses less information in order to initiate a scheduling. A. Discussion on Scheduling Adoption

Fig. 8. Waste capacity transported

Another critical issue is the capacity of waste transported with regards to each scheduling approach. In Figure 8, it is observed that for every value of top−𝑘 bins; dynamic scheduling achieves more waste collection capacity than the static scheduling. The low performance of the static scheduling is explained since the trucks collect waste from bins regardless of their bin capacity 𝑐. This means that a bin could be empty or half full (i.e., green or yellow) and it is collected in the case of static scheduling. However, this is not the case in dynamic scheduling where waste is collected only from the red full bins.

In order to adopt either dynamic or static scheduling approach an analyses of their strengths and weaknesses is necessary. Comparing the waste collection time and capacity transported; in Figure 7 and Figure 8, it is revealed the superiority of the dynamic scheduling approach. Instead, CPU overhead cost highlights static scheduling more cost efficient in strictly technical context; as presented in Figure 9. However, in the context of IoT-enabled Smart Cities the cost is not limited to strictly technical norms. This is because waste collection in Smart Cities has a strong social cost which has obviously higher impact in the society than the technical cost. Specifically, in the case of St. Petersburg; municipality authorities have a strong willing on investing in socially acceptable waste collection solutions, which means that in the context of Smart Cities dynamic scheduling is more likely to be adopted. VI. CONCLUSIONS AND FUTURE WORK We proposed a system architecture which incorporates a dynamic scheduling model as a solution to solid waste collection in Smart Cities. Dynamic scheduling is aware of which certain full bins capacity levels are reached. Moreover, dynamic scheduling enhances top−𝑘 queries with IoT realtime sensor and context information. The proposed dynamic scheduling model proved robust when compared with other static models studied in the literature. Dynamic scheduling evaluated on synthetic and real data from the city of St. Petersburg in Russia proved to be consistent in real time. Furthermore, the model and the techniques is the subject of ongoing discussion with St. Petersburg municipality authorities and waste management companies for deployment and experimentation purposes. Future research should be done in the areas of fully mobile depots (i.e., before waste transferred to the garbage tips). IoT offers a great promise for various Smart Cities scenarios, including smart waste collection, which demonstrated. ACKNOWLEDGEMENTS

Fig. 9. CPU overhead cost

Obviously waste collection time and capacity are main efforts in adopting a specific scheduling approach. However, CPU elapse time; with regards to overhead cost, should be analyzed to better understand the difference between dynamic and static scheduling. For the previous experimental case there are computed the CPU overhead cost between the two approaches, see Figure 9. It can be observed that the static scheduling has lower CPU overhead cost which is always smaller than the CPU overhead cost of dynamic scheduling approach. Specifically, dynamic scheduling CPU overhead cost is varied with respect to the value of top−𝑘 bins; thus the higher the 𝑘 value the more CPU overhead cost,

The research was carried out with the financial support of the Ministry of Education and Science of the Russian Federation under grant agreement #14.575.21.0058. REFERENCES [1]

[2]

M. Fazio, M. Paone, A. Puliafito and M. Villari, “Heterogeneous Sensors Become Homogeneous Things in Smart Cities”, 6th IEEE International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing (IMIS), pp. 775-780, Palermo, Italy, July, 2012. C. Balakrishna, “Enabling Technologies for Smart City Services and Applications”, 6th IEEE International Conference on Next

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

Generation Mobile Applications, Services and Technologies (NGMAST), pp. 223-227, Paris, France, September, 2012. T. Sanchez Lopez, D. C. Ranasinghe, M. Harrison and D. Mcfarlane, “Adding sense to the Internet of Things”, Personal and Ubiquitous Computing, Springer-Verlag, 2012, vol. 16 (3), pp. 291-308. A. J. Jara, P. Lopez, D. Fernandez, J. F. Castillo, M. A. Zamora, A. F. Skarmeta, “Mobile digcovery: discovering and interacting with the world through the Internet of Things”, Personal and Ubiquitous Computing, Springer-Verlag, 2014, vol. 18 (2), pp. 323-338. S. Suakanto, S. H. Supangkat, Suhardi and R. Saragih, “Smart City Dashboard for Integrating Various Data of Sensor Networks”, IEEE International Conference on ICT for Smart Society (ICISS), pp. 1-5, Jakarta, Indonesia, June, 2013. R. Carli, M. Dotoli, R. Pellegrino and L. Ranieri, “Measuring and Managing the Smartness of Cities: a Framework for Classifying Performance Indicators”, IEEE International Conference on Systems, Man, and Cybernetics (SMC), pp. 1288-1293, Manchester, UK, October, 2013. F. H. Priano and C. F. Guerra, “A Framework for Measuring Smart Cities”, In the Proceedings of the 15th Annual International Conference on Digital Government Research, DG.O ’14, pp.4454, Aguascalientes, Mexico, June, 2014. T. Nam and T. A. Pardo, “Smart City as Urban Innovation: Focusing on Management, Policy, and Context”, In the Proceedings of the 5th International Conference on Theory and Practice of Electronic Governance, ICEGOV ’11, pp. 185194,Tallinn, Estonia, September, 2011. T. Olivares, F. Royo and A. M. Ortiz, “An Experimental Testbed for Smart Cities Applications”, In the Proceedings of the 11 th ACM International Symposium on Mobility Management and Wireless Access, MobiWac ’13, pp. 115-118, Barcelona, Spain, November, 2013. R. Giffinger, C. Fertner, H, Kramar, R. Kalasek, N. PichlerMilanovic and E. Meijers, “Smart Cities: Ranking of European medium-sized cities”, Centre of Regional Science (SRF), Vienna University of Technology, Vienna, Austria, 2007, Online: http://www.smart-cities.eu [Cited on 27 November 2014]. C. Samaras, A. Vakali, M. Giatsoglou, D. Chatzakou and L. Angelis, “Requirements and architecture design principles for a smart city experiment with sensor and social networks integration”, In the Proceedings of the 17th Panhellenic Convference on Informatics, PCI ’13, pp. 327-334, Thessaloniki, Greece, September, 2013. D. Amagata, Y. Sasaki, T. Hara and S. Nishio, “A Robust Routing Method for Top-k Queries in Mobile Ad Hoc Networks”, 14th IEEE International Conference on Mobile Data Management (MDM), pp. 251-256, Milan, Italy, June, 2013. B. Chen, W. Liang and J. X. Yu, “Online Time Interval Top-k Queries in Wireless Sensor Networks”, 11 th IEEE International Conference on Mobile Data Management (MDM), pp. 251-256, Kansas City, Missouri, USA, May, 2010. A. Nadizadeha and H. H. Nasaba, “Solving the Dynamic Capacitated Location-Routing Problem with Fuzzy Demands by Hybrid Heuristic Algorithm”, European Journal of Operational Research, In Press Available Online, Elsevier, 2014. K. Buhrkal, A. Larsen and S. Ropke, “The Waste Collection Vehicle Routing Problem with Time Windows in a City Logistics Context”, Procedia Social and Behavioral Sciences, Elsevier, 2012, vol. 39, pp. 241-254. W. Juyoung, K.. Byung-In and K. Seongbae, “The rollon–rolloff waste collection vehicle routing problem with time windows”, European Journal of Operational Research, Elsevier, 2013, vol. 224 (3), pp. 466–476. M. Mes, “Using Simulation to Assess the Opportunities of Dynamic Waste Collection”, Use Cases of Discrete Event Simulation, Springer, 2012, pp. 277-307.

[18] I. Von Poser and A. R. Awad, “Optimal Routing for Solid Waste

[19] [20] [21]

[22]

[23]

[24] [25]

[26]

[27] [28]

[29] [30]

[31]

[32]

Collection in Cities by using Real Genetic Algorithm”, Information and Communication Technologies, ICTTA, IEEE, 2006, vol. 1, pp. 221–226. M. Mes, M. Schutten and A. P. Rivera, “Inventory routing for dynamic waste collection”, Beta conference, WP No. 431, Eindhoven, Netherlands, 2013. M. Reed, A. Yiannakou and R. Evering, “An ant colony algorithm for the multi-compartment vehicle routing problem”, Applied Soft Computing, Elsevier, 2014, vol. 15, pp. 169–176. Z. Zsigraiova, V. Semiao and F. Beijoco, “Operation Costs and Pollutant Emissions Reduction by Definition of new Collection Scheduling and Optimization of MSW Collection Routes using GIS. The Case Study of Barreiro, Portugal”, Waste Management, Elsevier, 2013, vol. 33, pp. 793-806. P. Milić and M. Jovanović, “The Advanced System for Dynamic Vehicle Routing in the Process of Waste Collection”, Facta Universitatis, Series: Mechanical Engineering, 2011, vol. 9 (1), pp. 127-136. T. Minh, T. V. Hoai, and T. T. N. Nguyet, “A Memetic Algorithm for Waste Collection Vehicle Routing Problem with Time Windows and Conflicts”, Computational Science and Its Applications, ICCSA, Lecture Notes in Computer Science, Springer, 2013, vol. 7971, pp. 485-499. J. Q. Li, D. Borenstein, P. B. Mirchandani, “Truck Scheduling for Solid Waste Collection in the City of Porto Alegre, Brazil”, Omega, Elsevier, 2008, vol. 36, pp. 1133-1149. P. T. R. Ramos, M. I. Gomes and A. P. B. Povoa, “Assessing and Improving Management Practices when Planning Packaging Waste Collection Systems”, Resources Conservation and Recycling, Elsevier, 2014, vol. 85, pp. 116-129. A. Stellingwerff, “Dynamic Waste Collection: Assessing the Usage of Dynamic Routing Methodologies”, Master Thesis, Industrial Engineering & Management, University of Twente, Twente Milieu, 2011. X. Bing, “Vehicle routing for the eco-efficient collection of household plastic waste”, Waste Management, Elsevier, 2014, vol. 34 (4), pp. 719-729. V. Hemmelmayr, K. F. Doerner, R. F. Hartl and S. Rath, “A heuristic solution method for node routing based solid waste collection problems”, Journal of Heuristics, Springer, 2013, vol. 19 (2), pp. 129–156. O. M. Johansson, “The effect of dynamic scheduling and routing in a solid waste management system”, Waste Management, Elsevier, 2006, vol. 26, pp. 875–885. T. Nuortio, J. Kytojoki, H. Niska and O. Braysy, “Improved route planning and scheduling of waste collection and transport”, Expert Systems with Applications, Elsevier, 2006, vol. 30, pp. 223–232. T. Anagnostopoulos and A. Zaslavsky, “Effective Waste Collection with Shortest Path Semi-Static and Dynamic Routing”, Proceedings of the 7th International Conference ruSMART, Internet of Things, Smart Spaces and Next Generation Networks and Systems, Lecture Notes in Computer Science, Springer, 2014, vol. 8638, pp 95-105. T. Anagnostopoulos, A. Zaslavsky, and A. Medvedev, “Robust Waste Collection exploiting Cost Efficiency of IoT potentiality in Smart Cities”, IEEE 1st International Conference on Recent Advances in Internet of Things (RIoT), [Accepted on February 4, 2015].