Method for optimization of sensor network placement

Abstract

Method for installing and / or optimizing a sensor network comprising a plurality of sensors, said method comprising the steps of receiving environmental data related to an area to be sensed, preferably an area containing a luminaire, and optionally constraint data related to the sensor network, identifying one or more suitable applications configured for using data sensed by one or more of the plurality of sensors of the sensor network, determining at least one characteristic or optimized characteristic of the sensor network based on the environmental data, the identified one or more applications, and optionally on the constraint data. A computer system or platform or infrastructure and an edge device corresponding to the method is also provided.

Description

[0001] METHOD FOR OPTIMIZATION OF SENSOR NETWORK PLACEMENT

[0002] FIELD OF INVENTION

[0003] The field of the invention relates to the field of sensor networks and more in particular, the determination and optimization of the placement of sensors in a sensor network. The field of the invention also relates to a computer system, platform or infrastructure therefor, as well as an edge device therefor.

[0004] BACKGROUND

[0005] Technological advancements in computing and telecommunication have drastically transformed the world. In particular, the growth of Internet of Things and cloud computing have been used to enhance the quality of services in cities. The integration of information, communication, and advanced sensors to manage city assets has lead to the concept of smart cities. In particular, sensor networks play a significant role in the collection of important information regarding the urban environment. Smart sensors are utilized in traffic control, curbside management, lighting, pollution control, garbage collection, energy management and monitoring, health, air quality, and food and water tracking such that smart cities improve sustainability and efficiency of various urban dynamics such as health, water, land, energy, etc.

[0006] SUMMARY

[0007] The amount of data that can be inferred from a sensor network and its use depends on multiple parameters, such as an amount of sensors that are present in the network, the position of the sensors, their environment, the types of sensors installed, etc. The installation of such a network of sensors also has a cost which may influence the choice of certain parameters. An object of embodiments of the invention is to provide a method that determines a sensor network to be installed and / or optimizes an already existing sensor network.

[0008] According to a first aspect of the invention, there is provided a method for installing and / or optimizing a sensor network comprising a plurality of sensors. Preferably, an existing luminaire network or traffic light network may be used for the installing and / or optimizing. The method comprises the steps of receiving environmental data related to an area to be sensed. Optionally, the method also comprises receiving constraint data related to the sensor network. The method comprises the steps of identifying one or more suitable applications configured for using data sensed by one or more of the plurality of sensors of the sensor network and of determining at least one characteristic or optimized characteristic of the sensor network based on the environmental data, the identified one or more applications and optionally on the optional constraint data. Preferably, the area includes a luminaire or a traffic light.

[0009] In other words, in exemplary embodiments the method may optimize a sensor network by using data sensed by the sensor network to identify one or more applications that are capable of using the sensed data in a meaningful way. The method also optimizes the sensor network by determining at least one characteristic of the sensor network, e.g. the position of sensors to be installed and / or optimized, their orientation, the amount thereof. This determination is based on environmental data related to the area to be sensed.

[0010] Optionally, additional data in the form of constraint data may be used for the determination.

[0011] In this way, a sensor network that better corresponds to the area to be sensed is obtained. The sensor network may not be installed yet. For example, city planners or the police may be interested in installing a sensor network in a given area, which does not comprise any sensor yet. Alternatively, the sensor network may be an already existing network, e.g., a sensor network that needs to be optimized, improved, or replaced. For example, the area to be sensed, which may or may not comprise the sensor network may comprise several sub-areas, with among them a first well illuminated and busy commercial area, with streets filled with traffic and parking spaces, a second calm and dark residential area which is safe and a third dark residential area where several crimes, e.g., burglaries, thefts or graffiti’s, have happened in the last years. In that situation, the method may identify applications for the first area such as a traffic monitoring application and a parking monitoring application, due to the presence of the road and the parking spaces thereof. The method may also identify an air quality and / or noise monitoring application in the first commercial area, since heavy traffic may induce a high level of noise and air pollution. A crime monitoring application may also be expected to be identified in the third residential area. The method may also determine, based on the environmental data and on the identified applications, that to install and / or optimize the sensor network, it may be interesting to install, add and / or replace some visible light cameras in the commercial area, e.g. install and / or add several cameras pointing at the road crossings to better monitor the traffic and the parking spaces, and to install and / or change the cameras in the residential areas by infrared cameras. It may also be determined to use a larger amount of cameras in the first commercial area and in the third residential area, where more control is needed, than in the calmer second residential area, where not much happens, and to install and / or add air quality sensors and noise sensors in the first commercial area, to better monitor the fluctuations of the air and noise pollutions levels generated by the heavy traffic.

[0012] Optionally, constraint data such as a budget or constrained position, which are “external” to the sensor network and the environment of the area, may also be used in the determination of the at least one characteristics of the sensor network. For example, the constraint data may comprise total budget and / or the price of each type of camera, such that users of the sensor network, such as city planners or the police, may know in advance what is achievable within an allocated budget. In another example, the constraint data may comprise constrained positions of the sensors. For example, the position of the sensors may be limited to some predetermined positions, e.g. on top of existing luminaires or traffic lights. In this way, the sensor network may also better correspond to external constraints.

[0013] Thus, the method allows optimizing and therefore improving the quality of the information that may be inferred from the sensor network such that the sensor data better corresponds to the area to be sensed and to the optional constraint data.

[0014] In a preferred embodiment, the sensor network may use an existing luminaire network or traffic light network. In an even more preferred embodiment, the area to be sensed may be an area including one or more luminaires of the luminaire network that is used or one or more traffic lights of the traffic light network that is used.

[0015] Luminaires, especially outdoor luminaires, have long been present worldwide to provide the lighting of cities, industrial plants, private or public parking lots, etc. Luminaires are often mounted on long poles or placed at a significant height such as to illuminate wider portions of an area such as a road or a parking space. By using these luminaires, advantage can be taken of the height of the luminaire and the pole they are often mounted on. This allows for example to capture images from an elevated point of view and to access information otherwise not easily obtained at ground level.

[0016] In a preferred embodiment, the step of identifying one or more applications configured for using data sensed by one or more of the plurality of sensors of the sensor network may comprise the selecting of one more suitable applications from a list of applications. For example, the sensor network may be installed or optimized in view of a defined purpose, e.g. an environmental purpose. In that case, the list of applications may comprise applications such as air quality monitoring, noise monitoring, littering monitoring, etc. , and the identified applications may be limited to these applications. In another example, the list of applications may comprise a list of the most commonly used applications.

[0017] In a preferred embodiment, as mentioned above, the sensor network may be an already existing sensor network comprising a plurality of already existing sensors. For example, the area to be sensed may already comprise a camera and an air quality sensor. The environmental data may then comprise environmental data sensed by one or more of the plurality of already existing sensors, e.g. by the camera and / or by the air quality sensor.

[0018] In that way, the method may include data sensed in situ, which may be more up to date than environmental data from available databases. Additionally, in this way, information on the characteristics of the already existing sensors is then available and up to date. For example, the camera may have moved, or its orientation may have been changed, e.g. due to vibration or due to maintenance and / or due to a change that has not been documented. Such orientation may cause the camera to point at a pedestrian sidewalk and at only one of the lanes of a road in the area, instead of pointing at the two lanes of a street. In this way, the characteristics to be improved, e.g., a change of orientation of the camera so that it points at the road and not at the sidewalks, are more easily determined.

[0019] In a preferred embodiment, the environmental data related to the area to be sensed may comprise any one or more of traffic data, such as vehicle / people counting data, ghost driver related data, traffic violation data, parking data, land use plans, air circulation data, noise maps, noise propagation maps, pollution data, weather data, crime data, data about one or more existing sensors if present in the area to be sensed, such as the position and / or type thereof, luminaire or traffic light data, such as position data and / or properties of a luminaire or traffic light if present in or in the vicinity of the area to be sensed, use type data, such a use as a terrace, a bicycle parking use, a (un)loading area use, etc., a user type data, such as use by children, animals, wheelchair users, elderly people, etc., objective related data related to environmental objectives for the area, such as less crime, less pollution, etc..

[0020] Traffic data may comprise information about traffic jams and delays, traffic intensities, rush-hour times and / or lanes, bridge openings, local speeds of passing vehicles, vehicle categories, road availability (whether the road is in good condition or whether road works are temporarily blocking some or all lanes of a road), safety (whether the road is sometimes slippery due to ice or is flooded due to rain). Land use plans data may comprise information related to land use of an area, policy zones, policy designations, regulation of the kinds and densities of activities which are acceptable and / or currently performed on particular lot of an area, e.g. residential, commercial, agricultural, industrial zonings, roadways, school zoning, parks, rivers and streams, etc. Crime data may include distributions of crimes in an area, their type, the number of occurrences of these crimes, the police presence in the area. In some cases, one or more sensors may already be present in the are to be sensed. In these cases, taking into account the data that can be obtained from these sensors may influence the way in which the sensor network is installed and / or optimized. Data about the one or more existing sensors may include the position of these sensors, their type, their orientation, their accuracy, etc. Luminaire data may include information related to the position of luminaires in the area, its height (as mentioned above, the height of the luminaire and the pole they are often mounted on may be taken advantage of), lighting distribution and intensity it provides in the area (whether the area is well lit, at what times, etc.). In this way, the method is adaptable to a broad range of uses and situations.

[0021] It should be noted that the environmental data related to the area to be sensed may be limited and may therefore not cover the entire area. For example, the data may originate from a combination of different databases, for example Geographic Information Systems (GIS) based maps, and some data may be missing in some sub-areas of the area, e.g. due to the source of data for these different subareas being different. Limited sets of sampled data points may still be used to estimate the value of a variable over a continuous spatial field, e.g. the area. Several spatial interpolation methods exists, such as Kriging, Trend Surface, Triangulated Irregular Network, Inverse Distance Weighted, etc., the Kriging interpolation method having the advantage of being based on the spatial arrangement of the empirical observations, rather than on a presumed model of spatial distribution, and generating estimates of the uncertainty surrounding each interpolated value.

[0022] In a preferred embodiment, the method may further comprise the step of receiving constraint data related to the sensor network. The determining of the at least one characteristic or optimized characteristic of the sensor network may then be further based on said constraint data.

[0023] Constraint data may comprise any one or more of: cost data related to the cost of the sensor network, sensor number data, such as a maximum or minimum number of sensors in the sensor network or a maximum or minimum number of different types of sensors in the sensor network, sensor type data, accuracy data, reliability data, sensor coverage data, sensor position data, such as preferred location data of a sensor to be added or to be relocated, a legal constraint such as a GDPR constraint or a no open source model constraint, application related data, such as a maximum or minimum number of applications to be used, resilience or weather condition related constraint data.

[0024] For example, the city may allocate a given budget for a sensor network aimed at fighting crime within a given area. This budget will often comprise all sorts of costs related to the sensor network, e.g. the cost of the personnel involved with the planning, of the purchase of the sensors, of the installation thereof, etc. By having this constraint within the method of optimization, a better solution for the needs of the user, that is better adapted to the budget, may be achieved. In addition, other constraints may alternatively or additionally be taken into account. Sensor number, type, accuracy, reliability and position data may comprise information related to the sensors that may be installed in the sensor network. Sensors may come in a wide variety of types and models, which vary in the data they can provide, in cost, in reliability, in the bandwidth they have, in the sensor coverage they can provide, in the temperature range at which they can operate. A cheap camera will not have the same quality of data as a more expensive, high-resolution camera. However, several cheap cameras may be bought for the same price as a more expensive one and depending on the use of the information provided by these cameras, it may well be that one option is unexpectedly better than the other. Reliability and accuracy may also be influenced by environmental factors, such as heavy rains, fog or humidity in general. Knowing the reliability and accuracy of the sensors and their dependence on external conditions may improve the installation and / or optimization of the sensor network. The number of sensors may also be limited. For example, a city may already have a stock of cameras ready to be installed, a fabricant may be limited in the amount of cameras of a given type that they can produce, or sensors may only be installed at a predetermined amount of locations, e.g. on luminaires or traffic lights. Constraint data may therefore allow for more “particularized” sensor networks.

[0025] The sensor network may be optimized in numerous ways. In a preferred embodiment, the at least one characteristic or optimized characteristic of the sensor network may comprise the identified one or more suitable applications. Additionally, the optimized characteristic may also be any one or more of the following: one or more suitable sensor types to be used in the sensor network, one or more suitable combination of different sensor types to be used in the sensor network, one or more suitable sensor positions, one or more suitable orientations of a sensor of the sensor network, a number of sensors to be included in the sensor network, one or more suitable processing models configured to process data sensed by the sensor network. The sensor type may comprise the type of sensed data or the type of sensor to sense said data, e.g. an image capturing means such as a visible light camera, an infrared light camera, an ultraviolet camera, a lidar sensor, an air quality sensor, a sound sensor such as a microphone, a thermometer, etc. The sensor type may also comprise the type of sensor within a given category, e.g. a high resolution camera, a low resolution camera, a high frame rate camera, or even the type within a list of models of a given list of options, e.g. camera of model M and brand B. Additionally or alternatively, the optimized characteristic may comprise one or more suitable positions and / or orientations of the sensors and / or the number of sensors such that the method provides with a sort of blueprint of the sensor network to be deployed or to be optimized. For example, one of the sensors, e.g. a camera in a first sub-area, may be determined to be moved into a second sub-area, while another sensor oriented at a pedestrian sidewalk may be determined to be reoriented, e.g. towards the road.

[0026] In a preferred embodiment, the determining of the at least one characteristic or optimized characteristic may further comprise the step of selecting sensor types to be included in the sensor network from a list of available sensors, based on the one or more identified applications, the environmental data and the optional constraint data. For example, the list of available sensors may be a list of the sensors that are already installed in the network. The list may also be a list of all types of sensors that can be installed in the network, for example because they are provided with the adequate attaching means, e.g. a luminaire attaching means. Alternatively, the owner and / or the user of the sensor network may already be in possession of a given number of sensor to be installed or may have contracts with some given manufacturer of sensors that may provide the user with predetermined models of sensors at a reduced rate. Alternatively or additionally, the determining may comprise the step of determining suitable positions for said selected sensors based on the one or more identified applications, the environmental data and the optional constraint data.

[0027] In a preferred embodiment, the one or more suitable applications may comprise any one or more of the following: an application related to the determining of traffic related data, such as a vehicle or pedestrian counting application, a ghost driver detection application, a traffic violation detection application; an application related to the determining of parking related data; an application related to the determining of air pollution data and / or sound pollution data and / or noise pollution data; an application related to the determining of crime related data; a use type detection application configured to detect one or more use types of said area, such as a curbside use, e.g. use as a terrace, a bicycle parking use, a (un)loading area use, etc.; a user type detection application configured to detect one or more user types in said area, such as a child, animal, wheelchair user, elderly people, etc.

[0028] Optionally, application priority rules may be used to determine which of the identified one or more suitable applications is to be used.

[0029] Optionally, the identified at least one application uses one or more models, and the method comprises configuring the one or more models based on sensed data.

[0030] The one or more models may comprise for example any one or more of the following: an object detection model, an object trajectory determining model, a model configured to determine if an object crosses a line, an object type determining model configured to determine a type of object. For example, a ghost driver detection application may use a vehicle detection model, a vehicle trajectory determining model and a model configured to determine if a vehicle crosses a line in a certain direction, wherein the latter will use the former for its determination.

[0031] The one or more models may be pre-trained models and / or the method may comprise training the one or more models based on data sensed by the sensor network, wherein optionally the one or more models are one or more neural networks models. For example, first the method may operate the application in a training mode for training the one or more models, whereupon the application may be used in an operational mode for the intended application.

[0032] In a further aspect of the invention, there is provided a method for the commissioning of a sensor network. The method comprises the steps of performing the method according to the first aspect, as well as the step of commissioning the sensor network based on the at least one characteristic of the sensor network that has been determined or optimized.

[0033] In other words the method of the first aspect is performed in order to obtain the at least one determined characteristic or optimized characteristic of a sensor network. This characteristic is then used as a blueprint to commission the sensor network. This may be used for example in the case where no sensor network exists yet.

[0034] In a third aspect of the invention, there is provided a method for optimizing a pre-existing sensor network. The method comprises the steps of performing the method according to the first aspect for optimizing the pre-existing sensor network, as well as the step of modifying the pre-existing sensor network based on the at least one determined optimized characteristic of the sensor network. Additionally or alternatively, the method comprises the steps of adding and / or removing one or more sensors to the pre-existing sensor network based on the characteristic of the sensor network that has been determined and / or optimized.

[0035] Modifying a pre-existing sensor network may comprise changing any kind of setting of a sensor of the pre-existing sensor network, for example changing a position and / or an orientation and / or resolution or frame rate of a camera, etc.

[0036] In other words the method of the first aspect is performed in order to obtain the at least one determined characteristic or optimized characteristic of a pre-existing sensor network. This characteristic is then used as a blueprint to modify the sensor network, e.g. modify the orientation of a sensor, modify the application used on the sensed data, and / or adding and / or removing one or more sensors.

[0037] It is noted that a sensor may be added to different types of edge devices, such as any one or more of the following: a luminaire, a bin, a sensor device, a street furniture, a charging station, a payment terminal, a parking terminal, a street sign, a traffic light, a telecommunication cabinet, a traffic surveillance terminal, a safety surveillance terminal, a water management terminal, a weather station, an energy metering terminal, an access lid in a pavement. The sensor system may be implemented in similar edge devices (e.g. a network of outdoor luminaires). Alternatively, the sensor system may be implemented in different edge devices. Existing structures ubiquitously present in cities may be used for hosting networks, limiting in this way the aesthetic impact of installing such networks. Structures having already access to the power grid are particularly interesting, while luminaires having typically a suitable height to capture all kinds of valuable data from sensors are further particularly suited as edge devices.

[0038] An edge device may be associated with one or more sensors, e.g. a sensor configured for obtaining a set of environmental data related to an event in the vicinity of the edge device, preferably at least two sensors configured for obtaining at least two sets of environmental data related to an event in the vicinity of the edge device. Optionally, the edge device may comprise a classification module configured to determine classification data of the event based on the at least one set, preferably at least two sets, of environmental data, and the classification data may be used to identify the one or more suitable applications and / or to determine the at least one characteristic. The edge device may be a luminaire comprising a light source and a drive means for driving the light source. The edge device may be further configured to control the drive means in function of an output of the at least one identified application. For example, when the application is a counting application, the light intensity and / or the light distribution and / or the color of the light emitted by the light source may be changed depending on the number of vehicles or persons counted within a predetermined period of time. In another example, the application is a traffic violation application and the light source may be controlled to blink or otherwise attract attention when a traffic violation is being detected.

[0039] Optionally, the at least one identified application is configured to use an output of the at least one sensor as an input of the application.

[0040] In a preferred embodiment, the at least one sensor comprises an image capturing means, a sound sensor and a radar sensor.

[0041] Optionally, the edge device is configured to configure the identified at least one application based on an output of the at least one sensor.

[0042] According to an exemplary embodiment, the sensor is selected from: an optical sensor such as a photodetector or an image sensor, a sound sensor, a radar such as a Doppler effect radar, a LIDAR, a humidity sensor, an air quality sensor, a temperature sensor, a motion sensor, an antenna such as a Bluetooth antenna for a Bluetooth sensor, an RF sensor, a metering device (e.g. a metering device for measuring the power consumption of a component of an edge device, more in particular a metering device for measuring the power consumption of a driver of a luminaire), a vibration sensor, a malfunctioning sensor e.g., a sensor for detecting the malfunctioning of a component of an edge device such as a current leakage detector for measuring current leaks in a driver of a luminaire), a measurement device for measuring a maintenance related parameter of a component of the edge device, an alarm device (e.g. a push button which a user can push in the event of an alarming situation). In this way, environmental data about an event in the vicinity of an edge device may be detected, e.g. characteristics (presence, absence, state, number, direction, speed, wearing mask or not) of objects like vehicles, street furniture, animals, persons, sub-parts of the edge device, or properties related to the environment (like weather (rain, fog, sun, wind), pollution, visibility, earth quake) or security related events (explosion, incident, gun shot, user alarm) in the vicinity of the edge device, maintenance related data or malfunctioning data of a component of the edge device. For example, in case of a crisis, the model may be configured to set one or more configuration parameters allowing to detect additional features like whether people are wearing mask or not, e.g. during a limited period of time.

[0043] According to an exemplary embodiment, the first and / or second sensor may be mounted in a housing of an edge device, e.g. a luminaire, in an orientable manner. An example of a suitable mounting structure is disclosed in WO 2019 / 243331 Al, in WO 2022 / 122755 Al and WO 2022 / 189601 Al in the name of the applicant which are included herein by reference. Such mounting structure may be used for arranging e.g. an optical sensor in the housing of an edge device. Other suitable mounting structures for mounting the first and / or second sensor are described in WO 2019 / 053259 Al, WO 2019 / 092273 Al, WO 2020 / 053342 Al, WO 2021 / 094612 Al, all of which are in the name of the applicant and included herein by reference. Although those patent specifications relate in particular to luminaire edge devices in which one or more sensors are provided, the skilled person understands that one or more sensors may be mounted in a similar way in another type of edge device.

[0044] According to an exemplary embodiment, the one or more edge devices are arranged in a mesh network. In this way edge devices can communicate directly with each other for a fast control. Reliability is also further improved due to redundant communication links. In particular, a fog device may also part of the mesh network. Preferably the edge communication means and optionally the fog communication means are configured to communicate through an IEEE 802.15.4 protocol.

[0045] Preferably, the edge devices are configured to transmit edge processed data to its associated fog device using a wireless personal area network (WPAN), preferably as defined in the IEEE 802.15.4 standard. According to another exemplary embodiment, the edge devices are configured to transmit edge processed data to its associated fog device through an LPWAN network, e.g. a LoRaWAN network or a SigFox network.

[0046] For example, the communication between the edge devices and its associated fog device may be based on a short range protocol such as IEEE 802.15.4 (e.g. Zigbee) or on a long range communication protocol such as LoRa wireless data communication technology. The network may be managed by the fog device or by a separate segment controller. In such a solution the edge device may only be capable of communicating through the short range communication protocol. However, it is also possible that some edge devices are capable of communicating both through a short-range and a long-range protocol. Also the fog device may be integrated with one of the edge devices, e.g. one of the luminaires of a group of luminaires could function as the fog device for a group of edge devices comprising the group of luminaires and possibly also other edge devices. Preferably, the fog devices are capable of communicating through a network comprising a cellular network with a central control system.

[0047] According to an exemplary embodiment, first processed data is obtained based on sensed data of a first sensor and second processed data is obtained based on data sensed by a second sensor, and measurement settings of the first sensor are adjusted based on the second processed data and / or measurement settings of the second sensor are adjusted based on the first processed data. In particular, variables of operation of a sensor like the frequency of a sound sensor, the light exposure or view field of a camera may be dynamically adjusted to benefit from the self-learning capabilities of the system. A measurement setting typically refers to a setting which affects the sensed environmental data obtained by the sensor whilst the set of rules used to process the sensed environmental data will influence the processing of the sensed data. Thus, according to exemplary embodiments not only the processing but also the measurements itself may be improved.

[0048] According to an exemplary embodiment, there is provided a communication means configured to receive data about an event in the one or more edge devices or in their vicinity from a central database and / or from the fog device. The first and / or second processing means may then be configured to use said data for processing the first and / or second sensed data and / or for training the first and / or second processing means. In this way more depth in the classification may be attained using additional information from a database. For instance, special vehicles like ambulances may be further identified using data from a dedicated database. Preferably the data in a database includes any one or more of the following: weather related information for the vicinity of the subset, traffic information for the vicinity of the subset, geo-coordinates of the edge devices of the subset, news and internet information, public transportation information, events (fairs, concerts, etc.) schedules, timing information (weekday, weekend, public holiday), public safety information, sanitary reports, security reports, road condition reports and cellphone data of cellphones in the vicinity of the subset. In this way, context data may be incorporated in the classification and / or training process for further accuracy.

[0049] According to an exemplary embodiment, the control means is configured to adjust measurement settings of the first and / or second sensor based on the received data. In this way, data from a database may be put to use for calibration of the sensors, increasing further the accuracy of the multi-sensor system. According to a further aspect of the invention, there is provided a computer system or platform or infrastructure for installing and / or optimizing a sensor network comprising a plurality of sensors preferably using an existing luminaire network. Said system or platform or infrastructure is configured for receiving environmental data related to an area to be sensed, preferably an area containing a luminaire, and optionally constraint data related to the sensor network, for identifying one or more suitable applications configured for using data sensed by one or more of the plurality of sensors of the sensor network and for determining at least one characteristic or optimized characteristic of the sensor network based on the environmental data, the identified one or more application, and optionally on the constraint data.

[0050] The skilled person will understand that the advantages that have been discussed above are still valid for this aspect of the invention.

[0051] BRIEF DESCRIPTION OF THE FIGURES

[0052] This and other aspects of the present invention will now be described in more detail, with reference to the appended drawings showing exemplary embodiments of the invention. Like numbers refer to like features throughout the drawings.

[0053] Fig. 1 A illustrates an area to be sensed by a sensor network in a city.

[0054] Fig. IB illustrates the steps of the method of the first, second and third aspects for installing and / or optimizing the sensor network in an area to be sensed.

[0055] Figs. 2A-D illustrate an area to be sensed of a city and examples of land use plans data (A), traffic and parking data (B), crime data, luminaire data and data about one or more existing sensors present in the area (C) and pollution data (D) that may be used as an environmental data received in the method.

[0056] Fig. 3 illustrates an example of the at least one characteristic of the sensor network that is determined.

[0057] Figures 4-7 illustrate an exemplary embodiment using an image capturing means and two sensors.

[0058] DESCRIPTION OF THE EMBODIMENTS Fig. 1 A illustrates an area to be sensed by a sensor network in a city.

[0059] Fig. 1A illustrates the road network of a portion of a city, where the portion is an area to be sensed 400. The city illustrated in Fig. 1A comprises a luminaire network, where each of the luminaires is represented as a black circle. Some of the luminaires are equipped with one or more sensors, represented as solid black disks, when there is only one sensor, and brick patterned disks, when there is a plurality of sensors, e.g. a combination of sensors. A zoom of the luminaire and sensor networks is illustrated on the right panel of the figure. Note that in this example, and in the following, the city comprises an already existing sensor network. However, the skilled person will understand that advantages of the invention may also be obtained when no already existing sensor network is present in the area to be sensed.

[0060] As visible from the right panel, the plurality of sensors 201a, 201b, 201c, 201d are connected (see solid line connecting the solid and dotted disks) and form a network 200. Similarly, the plurality of luminaires 251 are also connected (see dashed lines) and also form a network 250. Although here, the two networks have been represented as being independent, the skilled person will understand that the sensors 201 and the luminaires 251 may also be connected as one single network. Also, although not represented here, other configurations may be possible. For example, the luminaires 251 or the sensors 201 may be connected through fog devices or through a remote device.

[0061] The sensor network 200 may comprise different types of sensors. For example, sensors illustrated as solid black disks may be cameras, while sensors 201 illustrated as a dotted disk may be an air quality sensor. In the case of an existing sensor network 200, such as here, data sensed by the sensors 201 already in place may already be available prior to the installing and / or optimizing of the sensor network 200. In some cases, however, no sensed data may be available, for example if the sensor network has not been commissioned yet, or if the sensor is merely virtual, i.e. is planned to be installed but not yet installed nor active. In those cases, however, the type of data that will eventually be sensed is known, e.g. images for a camera, air pollutants concentrations for an air quality sensor. In the following, it will be referred to “sensed data” in a broad sense, even if no data has actually been sensed.

[0062] Fig. IB illustrates a system which may be used to perform the steps of the method of the first, second and third aspects for installing and / or optimizing the sensor network 200 in the area to be sensed 400. The method for installing and / or optimizing the sensor network 200 illustrated in Fig. 1A comprises the steps of receiving environmental data 101 related to the area to be sensed 400. As will be visible from Figs. 2A-D, the environmental data 101 may comprise a wide range of types of data. One or more suitable applications 301 are then identified, e.g. selected from a list of available applications, using data 210 sensed by one or more of the plurality of sensors 201a, 201b, 201c, 201d of the sensor network 200, i.e. using sensed data 210. As mentioned above, “sensed data” should be understood in a broad sense, even if no data has been actually sensed. Based on the environmental data 101 and the identified one or more applications 301, at least one characteristic or optimized characteristic 302 of the sensor network 200 is then determined. The characteristic determined by the method will be discussed in Fig. 3 in more detail.

[0063] Optionally the method also comprises receiving constraint data 102 related to the sensor network 200 and the determining is further based on the optional constraint data 102.

[0064] The sensor network 200 may use an existing luminaire network 250. As mentioned above in relation to Fig. 1A, sensors 201 and luminaires 251 may be connected as one single network, or sensors 201 may communicate through the luminaire network 250. The area to be sensed 400 may be, as illustrated in Fig. 1A, an area containing a luminaire 251 of the luminaire network 250 that is used. This will often be the case when the sensors 201 are mounted on or near luminaires which are often mounted on long poles or placed at a significant height.

[0065] This characteristic may be used as a blueprint to commission a sensor network to be installed. This may be used for example in the case where no sensor network exists yet, for example if the sensor network 200 is merely virtual. In the case of a pre-existing sensor network, the optimized characteristic or optimized characteristic may be used to modify the pre-existing network.

[0066] Figs. 2A-D illustrate the area to be sensed 400 illustrated in Fig. 1A and different examples of environmental data that can be received to perform the method for installing and / or optimizing a sensor network 200 such as the one illustrated in Fig. 1 A (and which has not been illustrated here in the sake of conciseness).

[0067] Environmental data may comprise land use plans data, such as illustrated in Fig. 2A. As visible in this figure, each portion of the area 400 may be assigned with different land uses. For example, the city may comprise several parks 401a (zigzag pattern), several residential areas 401b (vertical stripes pattern), an office area 401c (horizontal brick pattern), a commercial area 401d (checkerboard pattern), roadways 402 (solid black), agricultural zones, industrial zonings, school zoning, rivers and streams, etc. Land use may also comprise information about the times of openings of the commercial area, policy zones, policy designations, regulation of the kinds and densities of activities which are acceptable and / or currently performed in a particular lot of an area.

[0068] Environmental data may also comprise traffic data or parking data, such as illustrated in Fig. 2B. The traffic data may for example comprise the density of traffic of a given portion of the road. For example, the density of traffic and the traffic intensity of a given portion of the road corresponds to the density of black dots of the road, e.g. the roundabout 402a in the middle of the area is usually very busy, while the roads in the residential area 402b in the upper part of the city are usually much less busy. Although not represented here, the traffic data may comprise information about traffic jams and delays, rush-hour times and / or lanes, bridge openings, local speeds of passing vehicles, vehicle categories, road availability (whether the road is in good condition or whether road works are temporarily blocking some or all lanes of a road), safety (whether the road is sometimes slippery due to ice or is flooded due to rain), etc.

[0069] The parking data may comprise information on the location of the parking spaces, illustrated as black rectangles, and their availability, e.g. the black rectangles are filled when occupied 403a and hollow 403b when not. Although not represented here, the parking data may also comprise information on whether the parking space is a charging station for electric vehicles, the rates of the parking spaces, the parking policies, e.g. whether vehicles of a given type are allowed to park at given parking spaces at a given time, etc.

[0070] Environmental data may also comprise crime data, luminaire data and data about one or more existing sensors present in the area, such as illustrated in Fig. 2C. For example, the crime data may comprise distributions of crimes in an area and the number of occurrences of these crimes, e.g. white regions 404a may indicate that no crimes have been reported in the region, light thin vertical stripes 404b may indicate that a small amount of crimes have been reported, such as graffiti’s and burglaries, heavy vertical stripes 404c may indicate that major crimes have been reported, such as murders. Crime data may also comprise information on their type, on their precise location, on the police presence in the area.

[0071] As described above, one or more sensors 201 may already be present in the area to be sensed 400, e.g. the sensor network 200. Similarly, one or more luminaires 251 forming a luminaire network 250 may also be present in the area 400. For example, as in Fig. 1A, each of the luminaires 251 is represented as a black circle. Some of the luminaires 251 are equipped with one or more sensors 201, represented as solid black disks, when there is only one sensor, and brick patterned disks, when there is a plurality of sensors. Environmental data may then comprise data about the one or more existing sensors 201, for example the position of these sensors, their type, their orientation, their accuracy, their reliability, etc. Luminaire data may include information related to the position of luminaires 251 in the area, their height, a lighting distribution and intensity they provide in the area (whether the area is well lit, at what times, etc.).

[0072] Environmental data may comprise pollution data which is illustrated in Fig. 2D as more or less dotted areas, representing more or less polluted areas 405a, 405b, 405c. As illustrated in the left panel of Fig. 2D, data may be limited and may therefore not be complete over the whole area. This is the case for example on the diagonal stripes 405x portion of area 400. This may be due to the data originating from a combination of different databases, for example the data may be based on GIS based maps. Limited sets of sampled data points may still be used to estimate the value of a variable over a continuous spatial field, e.g. the area 400. Several spatial interpolation methods exist, such as Kriging, Trend Surface, Triangulated Irregular Network, Inverse Distance Weighted, etc., the Kriging interpolation method having the advantage of being based on the spatial arrangement of the empirical observations, rather than on a presumed model of spatial distribution, and generating estimates of the uncertainty surrounding each interpolated value. In the end, interpolated data may be obtained (see right panel of Fig. 2D) that compensates for the missing data. Although described in relation to pollution data, such interpolation methods can also be applied to other data.

[0073] Although some types of environmental data have been described in detail with respect to Figs. 2A- D, the skilled person will understand that other types of environmental data may also be used.

[0074] After receiving the environmental data 101 and on the basis of the environmental data 101 and the identified one or more application 301, at least one characteristic or optimized characteristic 302 of the sensor network 200 is determined. Fig. 3 illustrates some of the types of characteristics that may be determined by the method of Fig. IB.

[0075] For example, it may be determined that some of the sensors are not placed at optimal places. The optimized characteristic may therefore be to move the sensor to an already existing location. This is illustrated in Fig. 3 as movement 3021, where a combination of sensors (brick patterned disk) is determined to move close to the bottom right roundabout instead of the sensor that is already there. This movement 3021 may be explained by the heavy traffic and the poor parking availability in that area. The optimized characteristic may also be to move a sensor to a new location, where a luminaire is present. This is illustrated as movement 3022, where a camera (black disk) is moved to a new place where a luminaire is present. This may be explained by the presence of the commercial zone and the large number of parking spaces for which there is no sensor monitoring prior to the optimizing. Similarly, sensors may be moved to new locations where there is no existing sensor and no luminaire (illustrated as stars). This is illustrated with movement 3023, where a camera is moved at a crossing in the residential area in the upper part of the city. This movement may be explained by the need for better security in that area, where crime has been reported multiple times.

[0076] In addition, the optimized characteristic may also comprise the installation of new sensors. For example, it may be determined to increase the number of combined sensors 3024 (brick patterned disks) in new locations (starred) and to install a new type of sensor 3025 (triangle) at a new location, for example a noise sensor. As above, the placement of these new sensors may be due to a need for security, noise monitoring, etc.

[0077] The optimized characteristic may therefore comprise the one or more suitable sensor types to be used in the sensor network and / or the one or more suitable sensor positions of these new sensors and of the old sensors, if there are ones. Although not represented here, the optimized characteristic may comprise one or more suitable orientations of a sensor of the sensor network and / or the number of sensors to be included in the sensor network, but also the identified one or more suitable applications, modified parameters of the applications in use, modifying the network structure and connections, etc.

[0078] This characteristic may be used as a blueprint to commission a sensor network to be installed. This may be used for example in the case where no sensor network exists yet, for example if the sensor network illustrated in Figs. 1-3 is merely virtual. In the case of a pre-existing sensor network, the optimized characteristic or optimized characteristic may be used to modify the pre-existing network.

[0079] In an exemplary embodiment, a combination of an image capturing means, a sound sensor and a radar sensor, is used, wherein the at least one identified application is configured to use an output of the image capturing means, the sound sensor and the radar sensor as an input of the at least one application.

[0080] Multimodal data streams using an image capturing means and at least one sensor and deep learning architectures have several advantages over single -modality approaches for object detection, such as improved accuracy, robustness to noise and ambiguity, handling of complex scenarios and adaptability to diverse data. In a preferred embodiment, a feature-level fusion architecture for object detection based on YOLO features from images sensed by the image capturing means, spectral and rhythm features from sound sensed by a sound sensor and general descriptors from radar modalities sensed by a radar, may be used. After alignment, the sensed data may be combined with an attention mechanism into a single classification network.

[0081] The field of object detection is an integral part of various real-world applications, such as autonomous driving, surveillance, robotics or medical imaging. Accurately and efficiently detecting objects in a frame enables an easier interpretation of visual data which, in turn, facilitates an automated decision-making process for a wide variety of applications, such as object tracking, human-computer interaction or activity recognition.

[0082] Despite the progress brought forward by deep learning architectures in object detection tasks, there are still some challenges to overcome with regards to different real- world scenarios: deep learning models often struggle to handle occlusion or cluttered scenes, where objects of interest can be partially obscured or surrounded by other objects; a limited number of samples for training can lead to overfitting and poor generalization performance, since deep learning models typically require large amounts of labeled training data for effective learning; class imbalance can negatively impact the performance of these models, as they tend to be biased towards the majority classes; and achieving real-time object detection, especially in low-resource settings, can be extremely challenging since deep learning models with high accuracy may be computationally expensive. Consequently, multimodal approaches arose as a viable alternative to address these issues.

[0083] A multimodal approach using an image capturing means and at least other one sensor, refers to the integration of different information sources (for example visual, textual, audio, radar, among others) to enhance the learning process of a machine / deep learning architecture: by combining the information from different modalities and leveraging their complementary nature, multimodal learning models capture richer representations of the data and allow for an improvement in performance. These modalities can be processed independently and then fused at different levels, namely, early fusion (input level), late fusion (decision level) or intermediate fusion (feature level), and bring several advantages over unimodal approaches, such as robustness to noise and ambiguity or handling data variability. On the other hand, multimodal approaches may lead to increased model complexity and computational requirements, alongside difficulties in the process of data alignment, modality selection and choice of fusion strategy. Hence, in exemplary embodiments, an architecture for feature level fusion of image, sound and radar modalities in the context of an object detection application is proposed. A variety of multimodal fusion strategies may be used to combine visual, auditory, and other sensory modalities for object detection. For instance, depth information may be incorporated alongside RGB data for camerabased object detection, aiming to enhance the performance of existing single-shot detection networks by leveraging depth sensing, which can be easily obtained using depth cameras like Kinect or stereo setups. A network architecture that enables parameterization of the fusion point / layer for both RGB and depth information within the network layers, may be used. Preferably, RGB and depth information towards the mid to late layers of the network are fused.

[0084] Combining RGB with depth images may also be useful in the a-modal 3D detection problem, the task of detecting and localizing objects in a three-dimensional space, even when they are partially or completely occluded. A novel neural network called 3D-SSD may be used, which consists of two main components: hierarchical feature fusion and multi-layer prediction. The hierarchical feature fusion is where the appearance and geometric features from RGB and depth images are learned and fused at two stages using fusion blocks. The result of this fusion may then be used for object detection in the multi-layer prediction. The network may be evaluated on the SUN RGB-D and NYUv2 datasets. Multimodal machine learning may also be applied to sound detection. A method using a result-level fusion convolutional neural network (CNN) to detect drones and determine their presence in the surrounding environment, may be used. The features of sound signals, specifically Log-Mel spectrogram and Mel Frequency Cepstral Coefficient (MFCC) may be extracted and separately input into the networks. The results from the two networks are fused using evidence theory to obtain the final detection outcome.

[0085] Preferably, the data fusion takes advantage of the complementarity and redundancy of the various sources of data to exploit both intra-modality and inter-modality relationships. Data fusion approaches can go from a simple concatenation of data to the implementation of a self-attention fusion network , and a YOLO-MS framework with a Feature Interaction and Self- Attention Fusion Network (FISAFN) may be used as the backbone. The FISAFN comprises two modules: the Feature Interaction Module (FIM) and the Self- Attention Feature Fusion Module (SAFFM). The FIM extracts correlations between two modalities (visible and infrared images) to reconstruct the information components of each modality and enhance information exchange. The SAFFM establishes long-range information dependence between the modalities, filtering redundant features and enhancing complementary features. Such YOLO-MS framework may achieve better information richness of the fused features

[0086] In a preferred embodiment, an architecture for feature level fusion of image, sound and radar Modalities is used, preferably based on a YOLOv7 model for images, MFCC features extracted from sound signals, and radar descriptors in the context of object detection, preferably fused at the input stage for a deep classification attention network.

[0087] Such architecture allows to build a localizable, interoperable, cyber-safe, resilient, distributed, autonomous and connected urban infrastructure to solve common smart city challenges. The public lighting infrastructure may be used as host and hub for smart cities: taking advantage of an infrastructure established for decades, individual sensors, cameras or small cells can be integrated into streetlights or light poles to leverage future data-centric smart city applications. Consequently, in this scenario where great amounts of data are collected from several different sensory devices, the design and development of an edge computing architecture for data fusion are an opportunity, since multimodal learning can not only reduce detection error, but also grant more reliability to the multiple distributed sources, while fixing problematic data and extracting higher level information.

[0088] In an exemplary embodiment a model is used to provide the integration of the sound and radar modalities to the pre-existing image capturing means network. A number of non-consecutive minutes of captured video samples (with sound) and the corresponding radar descriptors from sensors placed on a lighting pole may be used. Figure 4 shows an example of an image frame captured. First the processing means performs an annotation process. In the example, the data set is composed by seven different classes of objects: person, bicycle, car, motorcycle, siren, bus and truck; to evaluate the performance of our multimodal approach, in Figure 7 the results obtained by means of a confusion matrix are presented, as well as the precision, accuracy, and recall metrics, defined by:

[0089] Precision = TP / (TP + FP) (1)

[0090] Recall = TP / (TP + FN) (2)

[0091] Accuracy = (TP + TN) / (TP + TN + FP + FN) (3) where TP, TN, FP, and FN correspond to true positives, true negatives, false positives, and false negatives, respectively.

[0092] Optionally YOLOv7 may be used due to the model’s capacity to extract features at different levels while maintaining a high-level understanding of the scene, and at the same time using an anchor box strategy that allows for the accurate localization of objects. The YOLOv7 model may be pre-trained. To extract features from the sound signals Mel Frequency Cepstral Coefficients (MFCC) may be used: the mel log-spectrogram analysis is a known approach in sound signal processing tasks; it brings several advantages such as perceptual alignment (mel scale approximates the nonlinear frequency response of the human auditory system), dynamic range compression (logarithmic compression reduces the dynamic range of the signal, allowing for the capture of both high and low- level acoustic features simultaneously) and frequency binning (aggregation of energy across adjacent frequency bands, allowing for identification of spectral patterns at different scales); as for the radar modality, descriptors, such as velocity, lifetime, azimuth and elevation, may be introduced as features and fed directly into an attention mechanism of the classification network.

[0093] Figure 5 illustrates an exemplary embodiment of the proposed architecture of the feature-level fusion process: for the image model, a pre-trained YOLOv7 model is loaded and evaluated on a subset of non-consecutive minutes samples, and when the training process is finished, features from the most relevant bounding boxes are extracted from the last layer before the detection unit - to extract the features from the middle layers of the YOLOv7 model, a PyTorch implementation may be used by incorporating the concept of a forward hook to extract features from a specific layer. The forward hook is a function that is executed during the forward pass of the model and is registered to a chosen layer. When the forward pass occurs, the hook captures the output of the specified layer and stores it as the ’’features” attribute of the model Instance - by leveraging this mechanism, our custom model allows for the parametrization of the layer from which features are extracted, providing flexibility in capturing and utilizing specific features of interest for various downstream tasks. As for the corresponding sound samples, MFCC may be extracted and on both set of inputs (image and sound) Principal Component Analysis (PCA) may be applied - effectively reducing the feature space while preserving the essential information. The application of PCA results in a significant reduction in feature size, allowing for more efficient processing, a vital attribute on an architecture meant to be deployed in edge devices and performed in real time. Next, the image and sound-reduced features, alongside the radar descriptors, may be introduced to a deep attention network that will be able to select and identify the most important features with regards to the classification process, while reutilizing the original YOLOv7 bounding boxes to deal with the predictions of object localization. In summary, this architecture intends for the sound and radar modalities to possibly correct miss- classifications from the YOLOv7 image model, and by fusing all modalities at a feature level (while guaranteeing the proper time alignment of all data) an enhanced discriminative power (ability to capture complex relationships between modalities), learning of complementary information and a robustness to missing modalities, is obtained, all while reducing model complexity and increasing computational efficiency. Experimental results evidence that this architecture for object detection represents a promising approach for addressing real-world challenges in smart city applications by combining image, sound and radar modalities. Figure 6 and Figure 7 show some preliminary results obtained with the proposed approach, without the usage of radar descriptors and with a two-layer neural network for classification, with evaluation metrics of: Accuracy = 0.83, Precision = 0.42, Recall = 0.40. On a first analysis, the application of PCA resulted in a significant reduction in feature size: for the YOEO features, the dimensionality was reduced from 3072000 to 750 components, resulting in a cumulative explained variance of 0.9593; similarly, the sound features were reduced from 38809 to 12 components, achieving a cumulative explained variance of 0.9996. This reduction in dimensionality not only helped to alleviate computational challenges but also aided in eliminating redundant information and focusing on the most informative features, ultimately enhancing the overall effectiveness of the multimodal architecture for object detection.

[0094] The results show that the approach is viable and the idea of enhancing the image model with sound and radar modalities is achievable: evaluation metrics show high accuracy alongside lower values for precision and recall, which indicates that the model is making correct overall predictions.

[0095] Multimodal learning offers several advantages over single-modality approaches: by combining information from different modalities, multimodal architectures demonstrate improved accuracy, robustness to noise and ambiguity, and adaptability to complex scenarios. These findings highlight the potential of multimodal learning to address the challenges faced by deep learning models, such as occlusion, limited training data, and class imbalance. In experimental results, the feature-level fusion of image and sound modalities lead to enhanced object detection performance compared to individual unimodal approaches: this suggests that the fusion of multiple modalities can effectively capture complementary information and improve the overall detection capabilities in real-world applications.

[0096] Optionally Y OEO’ s features from different layers may be used to observe the impact of each of them on the final results. Also, an attention mechanism to highlight the important regions and features in the fused representation may be used: this may provide some level of interpretability by indicating which modalities contribute more significantly to the final classification.

Claims

CLAIMS1. Method for installing and / or optimizing a sensor network (200) comprising a plurality of sensors (201a, 201b, 201c, 201d), said method comprising the steps of: receiving environmental data (101) related to an area (400) to be sensed, preferably an area containing a luminaire (251), and optionally constraint data (102) related to the sensor network (200); identifying one or more suitable applications configured for using data (210) sensed by one or more of the plurality of sensors of the sensor network (200); determining at least one characteristic or optimized characteristic (302) of the sensor network (200) based on the environmental data (101), the identified one or more applications (301), and optionally on the constraint data (102).

2. Method according to claim 1 , wherein the sensor network uses an existing luminaire network (250) or traffic light network.

3. Method according to claim 2, wherein the area (400) to be sensed is an area containing a luminaire (251) of the luminaire network (250) or a traffic light of the traffic light network.

4. Method according to any of the previous claims, wherein the identifying comprises selecting one or more suitable applications from a list of applications.

5. Method according to any one of the previous claims, wherein the sensor network is an already existing sensor network comprising a plurality of already existing sensors and wherein environmental data comprises environmental data sensed by one or more of the plurality of already existing sensors.

6. Method according to any of the previous claims, wherein the environmental data related to the area to be sensed comprises any one or more of: traffic data, such as vehicle / people counting data, ghost driver related data, traffic violation data, parking data, land use plans, air circulation data, noise maps, noise propagation maps, pollution data, weather data, crime data, data about one or more existing sensors if present in the area to be sensed, such as the position and / or type thereof, luminaire or traffic light data, such as position data and / or properties of a luminaire or traffic light if present in or in the vicinity of the area to be sensed, use type data, such a use as a terrace, a bicycle parking use, a (un)loading area use, etc., auser type data, such as use by children, animals, wheelchair users, elderly people, etc., objective related data related to environmental objectives for the area, such as less crime, less pollution, etc.

7. Method according to any of the previous claims, wherein the method further comprises the step of receiving constraint data related to the sensor network and wherein the determining of the at least one characteristic or optimized characteristic of the sensor network is further based on said constraint data.

8. Method according to any of the previous claims, wherein the constraint data comprises any one or more of: cost data related to the cost of the sensor network, sensor number data, such as a maximum or minimum number of sensors in the sensor network or a maximum or minimum number of different types of sensors in the sensor network, sensor type data, accuracy data, reliability data, sensor coverage data, sensor position data, such as preferred location data of a sensor to be added or to be relocated, a legal constraint such as a GDPR constraint or a no open source model constraint, application related data, such as a maximum or minimum number of applications to be used, resilience or weather condition related constraint data.

9. Method according to any of the previous claims, wherein the at least one characteristic or optimized characteristic of the sensor network comprises the identified one or more suitable applications and any one or more of the following: one or more suitable sensor types to be used in the sensor network, one or more suitable combination of different sensor types to be used in the sensor network, one or more suitable sensor positions, one or more suitable orientations of a sensor of the sensor network, a number of sensors to be included in the sensor network, one or more suitable processing models configured to process data sensed by the sensor network.

10. Method according to any one of the previous claims, wherein the determining of the at least one characteristic or optimized characteristic further comprises: selecting sensor types to be included in the sensor network from a list of available sensors, based on the one or more identified applications and on the environmental data and optionally on the constraint data; and / or determining suitable positions for said sensors based on the one or more identified applications and on the environmental data and optionally on the constraint data.

11. Method according to any one of the previous claims, wherein the one or more suitable applications comprise any one or more of: an application related to the determining of traffic related data, such as a vehicle or pedestrian counting application, a ghost driver detection application, a traffic violation detection application; an application related to the determining of parking related data; an application related to the determining of air pollution data and / or sound pollution data and / or noise pollution data; an application related to the determining of crime related data; a use type detection application configured to detect one or more use types of said area, such as a curbside use, e.g. use as a terrace, a bicycle parking use, a (un)loading area use, etc.; a user type detection application configured to detect one or more user types in said area, such as a child, animal, wheelchair user, elderly people, etc.

12. Method of any one of the previous claims, wherein application priority rules are used to determine which of the identified one or more suitable applications is to be used.

13. Method of any one of the previous claims, wherein the identified at least one application uses one or more models, and wherein the method comprises configuring the one or more models based on sensed data.

14. Method of the previous claim, wherein the one or more models comprise any one or more of the following: an object detection model, an object trajectory determining model, a model configured to determine if an object crosses a line, an object type determining model configured to determine a type of object.

15. Method of claim 13 or 14, wherein the one or more models are pre-trained models and / or wherein the method comprises training the one or more models based on data sensed by the sensor network, wherein optionally the one or more models are one or more neural networks models.

16. Method for commissioning of a sensor network, comprising the steps of: a. performing the method according to any of the previous claims; b. commissioning the sensor network based on the at least one determined characteristic or optimized characteristic of the sensor network.

17. Method for optimizing a pre-existing sensor network comprisinga. performing the method according to any of the claims 1-15 for optimizing the preexisting sensor network; b. modifying the pre-existing sensor network based on the at least one determined optimized characteristic of the sensor network and / or adding and / or removing one or more sensors to the pre-existing sensor network based on the determined optimized characteristic of the sensor network.

18. A computer program comprising computer-executable instructions to perform the method, when the program is run on a computer, of any one of the previous claims.

19. A digital data storage medium encoding a machine-executable program of instructions to perform any one of the steps of the method of any one of the claims 1-14.

20. A computer program product comprising computer-executable instructions for performing the method of any one of the claims 1-14, when the program is run on a computer.

21. A computer system or platform or infrastructure for installing and / or optimizing a sensor network (200) comprising a plurality of sensors (201a, 201b, 201c, 201d), preferably using an existing luminaire network (250), said system or platform or infrastructure being configured for: receiving environmental data (101) related to an area (400) to be sensed, preferably an area containing a luminaire, and optionally constraint data (102) related to the sensor network (200); identifying one or more suitable applications configured for using data (210) sensed by one or more of the plurality of sensors of the sensor network (200); determining at least one characteristic or optimized characteristic (302) of the sensor network based on the environmental data (101), the identified one or more applications (301), and optionally on the constraint data (102).

22. The computer system or platform or infrastructure according to claim 18, wherein the identifying comprises selecting one or more suitable applications from a list of applications.

23. An edge device, such as a luminaire, comprising at least one sensor, said edge device being configured- for emitting environmental data (101) related to an area (400) to be sensed, preferably an area containing a luminaire, based on data sensed by said at least one sensor;- for receiving at least one characteristic or optimized characteristic (302) of the sensor, which has been determined remotely based on the environmental data (101) and one or more identified applications (301) which use the environmental data; and- for reconfiguring the at least one sensor using the received at least one characteristic.

24. The edge device of the previous claim, wherein the edge device is a luminaire comprising a light source and a drive means for driving the light source.

25. The edge device of the previous claim, wherein the edge device is further configured to control the drive means in function of an output of the at least one identified application.

26. The edge device of any one of the claims 23-25, wherein the at least one identified application is configured to use an output of the at least one sensor as an input of the application.

27. The edge device of any of the claims 23-26, wherein the at least one sensor comprises an image capturing means, a sound sensor and a radar sensor.

28. The edge device of any of the claims 23-27, wherein the edge device is configured to configure the identified at least one application based on an output of the at least one sensor.

29. An edge device network comprising a plurality of edge devices, preferably according to any one of the claims 23-28, and a system according to claim 21 or 22, each edge device comprising a communication means, wherein the received environmental data is received at least partly through the communication means of the plurality of edge devices.

Images