This thesis outlines the use of measured data collected using the bGrid system to estimate the number of people in two rooms in the Microsoft office at Schiphol. The main objective is to derive a correlation that transforms the data into a specific number of people. The bGrid system consists of a network of sensor nodes that measure CO2 concentration, movement intensity, relative humidity, ambient temperature, infrared object temperature and sound intensity. These nodes are strategically placed throughout the building and are interconnected through a gateway that is also part of the bGrid system. The specific offices used had an equal area of roughly 16 m², a capacity for eight people and contained two bGrid sensor nodes each. Data was collected on two days, separated by one day, yielding 1396 minutes of data. Ground truth data was collected using direct observation from outside the offices as to not interfere with the measurements. Since the offices were of equal area and capacity, were both used for the same purpose and each contained two sensor nodes, the rooms were deemed equivalent and could therefore be combined. On the first day of observation, only room 1 could be observed but on the second day both rooms could be observed. The data from room 1 was used to synthesise occupancy models while the data from room 2 was used for validation. Two methods were used to model the occupancy. An approach based on Multiple Linear Regression used a combination of the movement intensity and CO2 concentration to achieve a performance of 93.49%, with performance being defined as the times the model returns a number of people that is within one person of the observed occupancy expressed as a percentage of the total number of iterations. This was achieved by first splitting the data based on the observed occupancy to derive specific regression coefficients for those levels of occupancy. These were used in a Simulink model that actively chose which regression coefficient to use based on the input. Secondly a method using a three-layer, fully connected neural network with a combination of hyperbolic tangent and leaky ReLu activations functions used the ambient temperature, relative humidity, movement intensity and CO2 data to achieve a performance of 93.86%. This was achieved using a network structure with three hidden layers with 19, 21 and 39 neurons respectively. These numbers were derived using the genetic algorithm to optimise for 43 iterations with the number of neurons per hidden layer and the learning rate as optimisation variables. The final result yielded two models that were able to estimate the number of people in room 2 within one person of the observed number of people, respectively 93.49% and 93.86% of the time. The resulting models could, when verified further with additional data, be used to aid HVAC systems with ambient temperature control; having a reliable metric for occupancy allows people to be added to the energy balance of a room, which in turn allows for the creation of more accurate models of the indoor thermal climate. These models would allow for model based temperature controllers as opposed to PID type controllers that are currently used in most office thermostats.

There is a high probability that communication between vehicles in a traffic network and between vehicles and traffic controllers will be the next major wave of technological innovation in traffic. The collection of data of communicating vehicles in traffic offers live information about the traffic conditions, which can be used to control traffic optimally. It is expected that in the coming years a new vehicle type will take part in traffic that is able to navigate through traffic without driver and communicates with traffic controllers to share its information and receive its optimal trajectory: the automated vehicle. The automated vehicle is different from the autonomous vehicle, which does not communicate with traffic controllers or other traffic.
The traffic mixture consisting of conventional, connected and automated vehicles will change gradually over time. Designing a traffic control strategy that performs well for traffic consisting of different shares of these three types of vehicles and requires only a limited amount of computational power will be a main challenge in traffic control the coming years.
This report proposes a traffic control strategy and assesses the influences of this strategy on the network wide traffic conditions using a traffic simulation. In this way, the research will provide insight in the effects of vehicle connectivity and automation on the network wide traffic conditions.

Based on the live traffic information from the communicating vehicles in a network, traffic can be controlled from two sides: the traffic signal control and the automated vehicle trajectory control. The traffic control method proposed in this report works as follows: first, the traffic controller collects all the information of connected and automated vehicles in the network to update the vehicle set. Locations and speeds of connected and automated vehicles can be obtained directly. Automated vehicles also provide the traffic controller of information of the non-communicating vehicles in their surroundings, detected by their sensors. Then, the virtual departure time of each vehicle will be determined, which is the time at which the vehicle would depart not hindered by other traffic or traffic signal and can be calculated using the kinematic laws. Also the corresponding virtual departure speed is determined for each vehicle. Then, starting from the first vehicle on the link, the expected departure time will be calculated, based on the virtual departure time and the traffic signal state. The expected departure times and speeds of the other vehicles on the link will be based on their virtual departure time, the expected departure time of the vehicles in front of the vehicle and the signal state. The virtual and expected departure times and speeds are used for both the signal control and the trajectory control.
This report presents a decentralized signal control to limit the computational power required in large traffic networks. This signal control is based on the original back-pressure algorithm, which compares the downstream traffic conditions with the upstream traffic conditions around an intersection and makes decisions for the signal control based on the difference in back-pressure of different movements on an intersection. The proposed signal control strategy takes the approaching and exiting vehicles within a certain range from the intersection into account. At the start of each time slot, a prediction is made for the pressure of a link on the intersection at the start of the next time slot, based on the current number of approaching and exiting vehicles on a link, their expected departure time from the intersection and the expected traffic demand. Based on the predicted pressures at every link that is connected with the intersection, the traffic controller decides whether it should switch the traffic signal at the start of the next time slot.
The proposed method for the trajectory control of the automated vehicles is based on the decisions of the traffic controller regarding the traffic signal and the expected departure times and speeds of the vehicles in the network. The main goal of the trajectory control is to use the green phases in an optimal way, i.e. maximizing the amount of departing vehicles per unit of green time. The trajectory control should limit accelerations and decelerations in the network as much as possible to limit emissions and traveller discomfort. Therefore, a third type of departure time and speed, the smooth departure time and speed, is introduced and compared with the expected time and speed of an automated vehicle to optimize its trajectory. The smooth departure time is the moment in time at which the vehicle would depart from the intersection if it would accelerate constantly to the expected departure speed over the total distance to the intersection. If the expected departure time is later in time than the smooth departure time, the vehicle will slow down and vice versa. If the smooth departure time is equal to the expected departure time, the vehicle will accelerate constantly towards the intersection.
The proposed control method is tested in a traffic simulation to obtain the effects of the method on the traffic conditions for different traffic scenarios. For each scenario the average traffic time delay, average green time, average and maximum queue length and average acceleration and acceleration time is retrieved from the simulation to assess the effects of the proposed methods on the traffic conditions. The simulated network consists of a two-directional main road with five signalized intersections with side roads. The traffic consists of only cars and other modes of transportation are neglected for simplicity. Different shares of connected and automated vehicles and different vehicle flows in the network are tested.

The simulation results show that the proposed signal control strategy leads to relatively short green times, which decrease the intersection throughput and the ability of adjusting the green time ratio between the two crossing roads to the traffic demand. Traffic on the main road encounters relatively longer green times and shorter queue lengths than traffic on the side roads. The throughput of the intersections does not satisfy the traffic demand and therefore, the first intersection will function as a bottleneck and lowers the traffic flow at the next intersections. Therefore, queues especially occur at the first intersection that vehicles encounter. The results show that the higher the penetration rate of communicating vehicles is, the better the traffic controller is able to adapt the signal sequence to the current traffic situation. The signal control requires several improvements in order to function optimally, but has the potency to be a suitable decentralized control method in a partly connected and automated environment.
Results regarding the proposed trajectory control show that a higher share of automated vehicles in the network results in a higher average acceleration in a relatively shorter total acceleration time, which is a result of the fluctuations in the vehicle speed due to the trajectory control. However, automated vehicles increase the average number of departing vehicles per unit of green time, which leads to shorter queues and less average travel time delay in the network.
Next to the proposed signal control strategy, an existing coordinated signal control strategy is implemented and tested in the simulation without connected or automated vehicles to compare the results with the proposed methods. Simulations of this green wave control result in larger average travel time delays and larger queue lengths due to relatively shorter green phases. The average acceleration, however, is relatively low, since the green wave strategy avoids vehicles to decelerate to complete standstill after they cross the first intersection of the traffic corridor.

This research shows that vehicle connectivity and automation offer the ability to improve network wide traffic conditions applying a decentralized traffic control method to control signalized intersections and automated vehicles. Although the proposed traffic control method is not optimized yet, it shows that low shares of connected and automated vehicles in traffic already make a significant difference in the network wide traffic conditions and that the signal sequences and automated vehicle trajectories can be even optimized if these shares get higher. Since the control strategy is decentralized, the required amount of computational power of a traffic controller is limited to the amount that the control of one single intersection requires and is not depending on the network size.
This research provides several recommendations for further research to improve the proposed strategy. Instead of taking only vehicles into account within a certain range from the intersection, which all have the same contribution to the back-pressure, an interesting alternative is to take all vehicles around an intersection into account for the back-pressure and add a certain weight to each vehicle based on the distance to the intersection. This weight can be used to determine the contribution of each vehicle to the back-pressure. Next, a threshold for switching and a dynamic slot time instead of a fixed one could be improvements to the proposed method and should be tested in further research.
The trajectory can be optimized especially regarding the vehicle accelerations and decelerations. It is recommended to avoid the occurring speed fluctuations in the proposed method by optimizing the acceleration rate that is used for automated vehicles to reach the desired trajectory.
It is recommended to test the proposed method in more realistic traffic simulations in further research to get more insight in the effects to real traffic. In that way the control method can be adapted to real traffic. The simulation of this research uses many assumptions regarding several parameters which require further research to optimize the performance of the control method. Last, before implementing the proposed method or a similar method in traffic, further research is required regarding the traffic safety. Especially the fact that automated vehicles are able to depart from intersections exactly at the moment the traffic signal switches to green could lead to unsafe situations, since other traffic users are not yet used to this behaviour of automated vehicles.

In this thesis, the effect of traffic light control on urban networks is investigated, with the main focus on the optimization algorithms used to solve the optimization problems resulting from a model predictive control approach. This will be done by using a macroscopic traffic flow model (the S-Model) to simulate the real-time traffic flows inside a network. A microscopic emission model (the VT-micro emission model) is further added to also include impact of the vehicle behaviour (accelerating, decelerating, driving at a constant speed, etc.) on the total amount of emission gasses that are exhausted by the vehicles inside the traffic network.

In this thesis the optimization algorithms will be investigated with respect to the reduction of the cost function as well as the computation time needed compute the (sub)optimal solution. The cost function can consist of the Total Time Spent (TTS) by the vehicles inside the network, the Total Emissions (TE) the vehicles exhaust while inside the network, or a combination of both.

We make use of Matlab for implementation of the models and the optimization algorithms and SUMO as a traffic simulator, which will be used to simulate a real-time traffic network.

The traffic flow model and the emission model are both non-smooth and non-convex, which in general would require the use of a global optimization algorithm together with multiple starting points. We also use a smoothening function on the models to work with a local optimization algorithm that works with derivatives of the cost function and both models. By using multiple starting points for this method, we hope to obtain similar results. The optimization algorithms that are implemented and investigated in this thesis are: the Genetic Algorithm (GA), the Simulated Annealing (SA) algorithm, the Pattern Search (PS) algorithm, and the Resilient backPROPagation (RPROP) algorithm.

To compare the results of the four optimization algorithms, four different scenarios are considered. The final results show that all control methods perform better than the FT controller. Overall the GA algorithm performs best without using a multi-start approach, with PS and SA having similar results. The RPROP method is either close to the other methods (0-2%) or quite far off (10-15%), depending on the scenario. When looking at computation time, PS is the fastest. It is twice as fast compared to the GA in most scenarios. SA however takes around fifteen times the amount of computation time compared to the GA. RPROP has varying results again compared to the GA algorithm. A better cost function for the TTS as well as a more optimized algorithm for the RPROP method can resolve both of these issues but future work might need to prove this.