WO2024065039A1

WO2024065039A1 - Methods and apparatus for communicating event data between sensor units

Info

Publication number: WO2024065039A1
Application number: PCT/CA2023/051270
Authority: WO
Inventors: Christian Parker
Original assignee: Individual
Current assignee: Individual
Priority date: 2022-09-27
Filing date: 2023-09-26
Publication date: 2024-04-04
Anticipated expiration: 2025-03-27
Also published as: CA3268599A1

Abstract

This application relates to monitoring the environment and events within the environment using multiple sensor units. Each sensor unit has a sensor array for monitoring the environment around the sensor unit, a transceiver for wireless communications with other sensor units, and a controller for processing data. The controller is configured to determine a baseline reference frame. Once the baseline reference frame is determined, the controller is configured to identify deviations from the baseline reference frame as an event, and to transmit data associated with the event via the transceiver. The sensor unit is also configured to form a local network with multiple other sensor units via the transceiver.

Description

Methods and Apparatus for Communicating Event Data between Sensor Units

TECHNICAL FIELD

[0001] The invention relates to sensor networks, and in particular to generating local networks of sensor units to monitor events within a common environment.

BACKGROUND

[0002] Typically, a network enabled sensor collects data and sends it to the cloud for processing. The remote processors in the cloud identify and analyze events using machine learning models. Alternatively, an edge computer is used to collect all data and process all data from the sensor network prior to sending it up to the cloud for archiving and further data analytics.

[0003] Communication network formation is completed for the purpose of sharing information. Wireless communication networks are configured to enable communication of devices over vast distances without the requirement of wiring the devices together. These networks are configured at the time of network creation to maximize throughput for all devices in range of the wireless network.

[0004] Edge computing results in silos of data with the edge computer typically getting model updates from the cloud and processing results on the edge independent of other sensors. This may result in limited learning potential for a network unless coordination is centralized by cloud resources and the data is post processed to extract further insights. These new insights then update the model in the cloud and are sent back down to the edge processed sensor device. This type of traditional architecture works well in areas that have strong communication technologies like Wi-Fi or cellular. In areas with sparse communication the learning models are limited by this architecture.

SUMMARY

[0005] In accordance with the invention, there is provided a sensor unit comprising: a sensor array of one or more sensors for monitoring the environment around the sensor unit; a transceiver configured to wirelessly receive and transmit data with other sensors in a temporary local network, the temporary local network being associated with a network I D, the network ID comprising a unique identifier and information identifying at least one sensor type; and a controller configured to process data and to exchange data with the sensor array and with the transceiver, the controller being configured: to determine a baseline reference frame using data received from the sensor array, and, once the baseline reference frame is determined, to identify deviations from the baseline reference frame as an event; wherein, in response to receiving a network ID, the controller is configured to increase the polling frequency of the sensors in the sensor array of the at least one sensor type identified in the network ID; and wherein, in response to detecting an event using one or more of the sensors in the sensor array, the controller is configured to transmit sensor data associated with the detected event using a temporary local network associated with a said network ID, wherein, where a live temporary local network has been initiated with a network ID identifying a sensor type corresponding to the at least one sensor used to identify the event, the sensor unit is configured to transmit the sensor data via the live temporary local network, and wherein, where no live local temporary network has been initiated with a network ID identifying a sensor type corresponding to the at least one sensor used to identify the event, the sensor unit is configured to initiate a temporary local network by creating a network ID and sending the created network ID to other sensor units in the area.

[0006] A sensor unit may periodically request other sensor units in the area to send any network IDs that they have initiated. A sensor unit may automatically transmit a network ID after it has been created. A sensor unit may only transmit network IDs corresponding to live temporary networks (e.g., initiated or established temporary networks, but not expired temporary networks). A network ID may be a temporary network ID.

[0007] When a sensor unit has initiated a temporary local network, the sensor unit may be configured to receive sensor data associated with an event associated with the network ID. [0008] The baseline reference frame may be a baseline of sensor array data over the course of a cycle period. The baseline reference frame may be determined using data received from the sensor array over multiple cycles. The baseline reference frame may be determined using data received from the sensor array a predetermined period of time. The baseline reference frame may be determined using data received from the sensor array while the sensor unit is stationary.

[0009] The unique component of the network ID may comprise time information relating to when the event was detected.

[0010] The sensor unit may be configured to determine the baseline reference frame partially based on baseline reference frame data received from another sensor unit located in the vicinity.

[0011] A sensor unit may collate information from multiple detected events into an event chain. The information may come from its own sensor array and/or from other sensor units. [0012] Sensor units may transmit and receive information directly with other sensor units (e.g., within an intermediate electronic device). The sensor units may communicate using wired and/or wireless channels.

[0013] A vicinity may be a predetermined area. For example, the vicinity of a sensor unit may be an area within a predetermined radius of the sensor unit.

[0014] The sensor unit may be configured to communicate with a local group of wirelessly connected sensor units via the transceiver, wherein the local group forms a low power wireless mesh network.

[0015] The sensor unit may be configured to periodically exchange baseline reference frame data with other sensor units in a local network, and to refine the baseline reference frame using the received baseline reference frame data.

[0016] The sensor unit may be configured to associate deviations from the baseline reference frame with a unique event identifier and a timestamp. The timestamp may correspond to the time or time period at which the deviation occurred.

[0017] The controller may be configured, in response to receiving event data comprising a timestamp and a unique event identifier from another sensor unit, wherein the received timestamp is within a predetermined period before an identified deviation detected by the sensor array, to associate the identified deviation with the received unique event identifier, so that the unique event identifier is associated with an event chain comprising information from multiple sensor units. This allows the sensor units to collate information on an event from multiple sensor units.

[0018] The controller may be configured to, in response to receiving event data comprising sensor data and a unique event identifier from another sensor unit, wherein the received sensor data is sufficiently similar to sensor data associated with an identified deviation detected by the sensor array, associate the identified deviation with the received unique event identifier, so that the unique event identifier is associated with an event chain comprising information from multiple sensor units.

[0019] The controller may be configured to associate two events detected at different sensor units when the events have sufficiently similar sensor signatures.

[0020] The controller may be configured to record an image of the environment, in response to: receiving a unique event identifier and timestamp corresponding to an event from another sensor unit; and, detecting an identified deviation within a predetermined period of the received timestamp.

[0021] The controller may be configured to change the mode of the sensor unit in response to receiving event data from another sensor unit in the local network. For example, the controller may change the mode from a passive mode to an active mode in response to another sensor unit determining that an event has occurred. Compared to the passive mode, the active mode may comprise taking sensor readings more frequently. The passive mode may still correspond to periodically taking readings (i.e., not an off or stand-by mode).

[0022] Each of the sensor units in the local network may have a common environment. E.g., they may all be outside, or all be inside in the same building. A sensor unit outside and one inside would not typically form part of the same local network.

[0023] The sensor array may comprise one of more of: a temperature sensor; a humidity sensor; and an ambient light sensor.

[0024] Polling may correspond to an attempt to record an event. Polling may comprise turning a sensor on, taking a measurement, and recording the results in memory. When the device is in passive mode, the polling frequency will be relatively slow (e.g., 0.5-10 minutes, maybe less, maybe more). In an active mode, the sensor unit may start more frequently accessing and storing sensor data (e.g., 10-20 times/second) for a period of time equal to a predetermined event time period threshold (e.g., 10-15 minutes or until the event ends). By doing this the sensor unit is primed to detect an event, and the start of the event can be recorded.

[0025] The sensor array may comprise one or more of: wind speed sensor; a wind direction sensor; a gas detector; and a height sensor. The height sensor may be configured to determine the height of a floor below the sensor or a ceiling above the sensor. The floor may correspond to the top of any dense material (solid and/or liquid) below the sensor. For example, it might include the tops of growing plants in a field, or the surface of water (e.g., to measure tides). Likewise, the ceiling sensor may measure the height of the underside of a canopy of trees.

[0026] The sensor unit may be configured to decide whether or not to separate from an established local network connection with another sensor unit based on the degree of similarity between the baseline reference frames of the sensor unit and the other sensor unit. This allows the local network continually to adapt as the sensed conditions provide more information on the environment.

[0027] The sensor unit may be configured to initiate communications with other sensor units in the area in response to determining that the position of the sensor unit has changed. This may allow the sensor units to initiate a local network connection without an explicit instruction from the user. That is, the sensor unit will know that the environment may have changed as it has been moved so, in response, it seeks to form a local network with other units in the new environment. Other embodiments may allow the user to manually initiate communications.

[0028] The sensor unit may be configured to take an image and transmit the image in response to detecting that the sensor unit is being moved. This may help identify theft of the unit.

[0029] The sensor unit may be configured to transmit event data once every cycle.

[0030] The sensor unit may be configured, in response to identifying the same deviation over multiple cycles: to update the baseline reference frame of the sensor unit; to redetermine the degree of similarity of the determined baseline reference frames of the sensor unit and baseline reference frames of the multiple other sensor units; and to stay or leave the local network based on the degree of similarity. [0031] The sensor unit may comprise multiple transceivers configured to receive and transmit data using multiple different communication protocols.

[0032] The local network may be a hybrid network in which sensor units are configured to communicate using multiple different network protocols.

[0033] According to a further aspect, there is provided an assembly of sensor units as described herein, wherein the sensor units are wirelessly connected in the local network. [0034] The sensor units may each have one or more sensors for measuring the same environmental parameter.

[0035] The assembly may be configured to determine an event reference frame, the event reference frame being a collection of consistent deviations of sensor data recorded by multiple sensor units across multiple identified events.

[0036] According to a further aspect, there is provided a method of forming a local network between a plurality of sensor units, each sensor unit comprising: a sensor array of one or more sensors for monitoring the environment around the sensor tower; a transceiver configured to wirelessly receive and transmit data with other sensors in a local network; and a controller configured to receive and process data from the sensor array, and to receive, transmit and process data via the transceiver; wherein the method comprises each sensor unit: determining a baseline reference frame using data received from the sensor array; once the baseline reference frame is determined, identifying deviations from the baseline reference frame as an event, and transmitting data associated with the event via the transceiver; and in response to receiving a network ID, increasing the polling frequency of the sensors in the sensor array of the at least one sensor type identified in the network ID; in response to detecting an event using one or more of the sensors in the sensor array, transmitting sensor data associated with the detected event using a temporary local network associated with a said network ID, wherein, where a live temporary local network has been initiated with a network ID identifying a sensor type corresponding to the at least one sensor used to identify the event, the sensor data is transmitted via the live temporary local network, and wherein, where no live local temporary network has been initiated with a network I D identifying a sensor type corresponding to the at least one sensor used to identify the event, initiating a temporary local network by creating a network ID and sending the created network ID to other sensor units in the area.

[0037] Once an event has been identified the sensor unit may transmit information on the type of sensor that has detected a deviation.

[0038] Once a transceiver receives information on the type of sensor, sensor units having sensors of the same type may form a network. Sensor units may form a local network associated with a network ID.

[0039] In the context of this disclosure, initiating a network comprises a single sensor unit creating a network ID and enabling the transmission of data associated with that network ID. Establishing a network comprises multiple sensor units connecting to each other using a created network ID and exchanging data with each other. An initiated network or an established network may be considered to be a live network. A live network may expire after a predetermined period of time following initiation, or following the last event data being associated with the associated network ID.

[0040] A root node may accept child nodes in the network through a network joining procedure.

[0041] All transceivers in range of the root transceiver that have sensors of the same type may move from a sleep/standby state into a data ready state which will include an increased frequency of sensor data polling.

[0042] Once a second transceiver has identified an event matching the identified sensor type identifier, the second transceiver may connect to the root network as a child node.

[0043] The sensor unit which is the root node may collate all the information received via a said local network. The sensor unit may store the information relating to an event in association with the network ID. The network ID may act as an event identifier. A child node may transmit information via a said local network directly to the root node, or via one or more intermediate nodes. [0044] Each sensor unit, on the detection of an event, in the absence of a matching wireless network may create their own wireless network to accept connections. Wireless networks may be removed after event threshold has been met. E.g., an event threshold may be a predetermined period of time without any further events being detected by the sensor units.

[0045] Each network may be configured to identify itself a root node. Each network may be configured to identify itself a child node.

[0046] Each sensor unit, in response to detecting an event when no live local temporary network has been initiated with a network ID identifying a sensor type corresponding to the at least one sensor used to identify the event, may be configured to identify itself as a root node.

[0047] Each sensor unit, in response to detecting an event that contains more sensor events than the currently available network broadcast sensor identifiers may elect to become the root node for then network and update the network broadcast information to contain all event identifiers.

[0048] Each sensor unit, in response to detecting an event occurring on fewer sensors than the currently available networks indicate through their non unique identifiers will join the root node and network with more active sensor events as a child node.

[0049] Each sensor unit may connect to a network only after validating a unique application specific publicly available key against a private key held on the sensor unit to validate authenticity of the root node as a root node that belongs to the asset owner of the sensor unit. In this way, co-located assets can be monitored without the sharing of critical operational data between network nodes not intended or allowed to be connected. This public private key exchange will also be used to facilitate network encryption.

[0050] The method of generating the unique identifier may include a memorized secret and a number of other types of cryptographic keys, including but not limited to a symmetric key or a private key. And the method may incorporate a time-based authentication protocol.

[0051] Each sensor unit may be a stand-alone unit capable of operating independently from other sensor units. The components of each sensor unit (e.g., including the sensor array, controller and transceiver) may be physically connected together. The components of each sensor unit may be housed within a housing such that the sensor unit can be lifted and transported as a single physical object.

[0052] According to a further aspect of the present disclosure, there is provided a sensor unit comprising: a sensor array of one or more sensors for monitoring the environment around the sensor unit; a transceiver configured to wirelessly receive and transmit data with other sensors in a local network; and a controller configured to receive and process data from the sensor array, and to receive, transmit and process data via the transceiver; wherein the controller is configured to determine a baseline reference frame, the baseline reference frame being a baseline of sensor array data over the course of a cycle period, the baseline reference frame being determined using data received from the sensor array over multiple cycles while the sensor unit is stationary; wherein, once the baseline reference frame is determined, the controller is configured to identify deviations from the baseline reference frame as an event, and to transmit data associated with the event via the transceiver; and wherein the sensor unit is configured to form a local network with multiple other sensor units via the transceiver based on geographical proximity and a degree of similarity of the determined baseline reference frames of the sensor unit and baseline reference frames of the multiple other sensor units.

[0053] According to a further aspect, there is provided a method of forming a local network between a plurality of sensor units, each sensor unit comprising: a sensor array of one or more sensors for monitoring the environment around the sensor tower; a transceiver configured to wirelessly receive and transmit data with other sensors in a local network; and a controller configured to receive and process data from the sensor array, and to receive, transmit and process data via the transceiver; wherein the method comprises each sensor unit: determining a baseline reference frame, the baseline reference frame being a baseline of sensor array data over the course of a cycle period, the baseline reference frame being determined using data received from the sensor array over multiple cycles while the sensor unit is stationary; once the baseline reference frame is determined, identifying deviations from the baseline reference frame as an event, and to transmit data associated with the event via the transceiver; and forming the local network with the other sensor units via the transceiver based on geographical proximity and a degree of similarity of the determined baseline reference frames of the sensor unit and baseline reference frames of the multiple other sensor units.

[0054] The sensor unit may be less than 2 meters tall. The sensor unit may be transportable (e.g., less than 25kg). The sensor unit may be battery powered. The sensor unit may comprise an onboard power generator (e.g., solar panels or wind turbine). The sensor unit may not require mains power to operate.

[0055] A transportable sensor unit is generally not designed to be worn on the users clothing. A transportable sensor unit may be designed to be moved from one location to another then remain stationary while in use. A transportable sensor unit may be greater than 1 kg. A transportable sensor unit may have a battery life greater than 24 hours. A transportable sensor unit may be connected to other fixed or temporary power sources. A transportable sensor unit may control other fixed or temporary electronic devices.

[0056] Audio sensors may be directional (e.g., 6 directions 60 degrees apart).

[0057] The creation of reference frames may be established using a distributed ledger and/or a blockchain technology. Distributed ledgers use independent computers (or nodes) to record, share and synchronize transactions in their respective electronic ledgers (instead of keeping data centralized as in a traditional ledger). Blockchain organizes data into blocks, which are chained together in an append only mode.

[0058] Network protocols may include any Wi-Fi, ethernet and/or Bluetooth™.

[0059] Mesh network protocols may include one or more of the following: mesh networks. Some of these include: Associativity-Based Routing (ABR); AODV (Ad hoc On-Demand Distance Vector); B.A.T.M.A.N. (Better Approach To Mobile Adhoc Networking); Babel (protocol) (a distance-vector routing protocol for IPv6 and IPv4 with fast convergence properties); Dynamic Nix-Vector Routing|DNVR; DSDV (Destination-Sequenced Distance-Vector Routing); DSR (Dynamic Source Routing); HSLS (Hazy-Sighted Link State); HWMP (Hybrid Wireless Mesh Protocol, the default mandatory routing protocol of IEEE 802.11s); Infrastructure Wireless Mesh Protocol (IWMP) for Infrastructure Mesh Networks by GRECO UFPB-Brazil; ODMRP (On-Demand Multicast Routing Protocol); OLSR (Optimized Link State Routing protocol); OORP (OrderOne Routing Protocol) (OrderOne Networks Routing Protocol); OSPF (Open Shortest Path First Routing); Routing Protocol for Low-Power and Lossy Networks (IETF ROLL RPL protocol, RFC 6550); PWRP (Predictive Wireless Routing Protocol); TORA (Temporally-Ordered Routing Algorithm); and ZRP (Zone Routing Protocol).

[0060] The methods and systems may employ artificial intelligence (Al) techniques such as machine learning and iterative learning. Examples of such techniques include, but are not limited to, expert systems, case-based reasoning, Bayesian networks, behavior-based Al, neural networks, fuzzy systems, evolutionary computation (e.g., genetic algorithms), swarm intelligence (e.g., ant algorithms), and hybrid intelligent systems (e.g., Expert inference rules generated through a neural network or production rules from statistical learning). The methods and systems may use reinforcement learning, deep neural networks and/or recurrent neural networks.

[0061] The machine learning may use supervised or unsupervised learning which comprises learning a function that maps an input to an output based on example inputoutput pairs. It may involve inferring a function from labeled training data consisting of a set of training examples. In supervised learning, each example is a pair consisting of an input object (e.g., measured sensor data) and an output value (e.g., identified event). A supervised learning algorithm may be configured to analyze the training data and produces an inferred function, which can be used for mapping new examples. Unsupervised learning could be used to identify the sensor data and find events in the data.

[0062] The controller may comprise a processor and memory. The memory may store computer program code. The processor may comprise, for example, a graphics processing unit, a central processing unit, a microprocessor, an application-specific integrated circuit or ASIC or a multicore processor. The memory may comprise, for example, flash memory, a hard-drive, volatile memory. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Various objects, features and advantages of the invention will be apparent from the following description of particular embodiments of the invention, as illustrated in the accompanying drawings. The drawings are not necessarily to scale, emphasis instead being placed upon illustrating the principles of various embodiments of the invention. Similar reference numerals indicate similar components.

Figure 1 is a schematic diagram of a sensor unit.

Figure 2 is a schematic plan view of several networked sensor units within a mining environment.

Figure 3 is a schematic plan view of several networked sensor units within a agricultural environment.

Figure 4 is a flow chart showing how the sensor unit interacts with other sensor units within range to establish ad-hoc networks for communicating data relating to an event.

DETAILED DESCRIPTION

Introduction

[0064] The present technology relates to environmental monitoring using multiple stationary stand-alone sensor units. Each sensor unit in the system may have sensors configured to monitor the same environmental parameter (e.g., temperature, humidity, noise level, noise spectrum) to allow comparison between the different sensor units. Sensor units may be set up around an industrial or agricultural site to monitor the site. As part of this monitoring, the system is configured to determine a baseline of what is expected to happen (e.g., at particular times), and to identify deviations from that baseline as being an event. Identifying events allows users of the system to recognise and understand unusual occurrences more easily.

[0065] The present technology addresses issues with event data being centrally processed, away from the local network in a central cloud server and being transferred over a cellular or satellite connection.

[0066] The present technology enables network enabled sensor units to self-organize and self-train machine learning models without the requirement for cloud resources. That is, the present technology enables the creation of distributed reference frames based on multiple sensor units detecting events using the same type of sensors (e.g., multiple sensors identifying an audio event and then collating data). All processing to establish reference frames and identify events may occur using devices connected to the local network.

[0067] Various aspects of the invention will now be described with reference to the figures. For the purposes of illustration, components depicted in the figures are not necessarily drawn to scale. Instead, emphasis is placed on highlighting the various contributions of the components to the functionality of various aspects of the invention. A number of possible alternative features are introduced during the course of this description. It is to be understood that, according to the knowledge and judgment of persons skilled in the art, such alternative features may be substituted in various combinations to arrive at different embodiments of the present invention.

Vehicle Access Detection

[0068] Figure 1 shows a first embodiment of a sensor unit 100. In this case, the sensor unit comprises a sensor array 120, in this case having three sensors for monitoring the environment around the sensor unit. In this embodiment the sensor array comprises a thermometer or temperature sensor 121 for determining the temperature of the environment, a microphone 122 for detecting sound (or audio), and an image sensor 123 for recording optical images of the environment around the sensor unit.

[0069] The sensor unit also comprises a transceiver 102 configured to wirelessly receive and transmit data with other sensor units in a local network; and a controller 101 configured to receive and process data from the sensor array, and to receive, transmit and process data via the transceiver.

[0070] Figure 2 shows a mine site environment in which multiple sensor units 200a-d (like those shown in figure 1) have been installed at strategic places around the mine site. These include a first sensor unit 200a being at the entrance to the mine site, a second sensor unit near the crusher 200b, a third sensor unit 200c on the road between the quarry and the entrance, and a fourth sensor unit 200d at the entrance to the quarry.

[0071] In this embodiment, each sensor unit is configured to determine a baseline reference frame, the baseline reference frame relating to a baseline in cyclic changes in the sensor data received from the sensor array over multiple cycles while the sensor unit is stationary. That is, the baseline reference frame in this example may be considered to be a time dependent prediction of what the expected sensor data should be over a cycle period. This allows the controller to determine if a particular sensor reading is as expected or a deviation by comparing the sensor reading to the corresponding baseline reference frame value (or range of values) for the corresponding point in time in the temporal cycle. It will be appreciated that it typically would take a sensor unit several cycles to determine the baseline reference frame.

[0072] In this case, the baseline reference frame will include data from all three sensors in that unit including temperature data, sound data and image data as a function of time. In other embodiments, each sensor unit may determine its own baseline reference frame. In this embodiment, the image data is configured to detect motion or changes in the image (e.g., resulting from a passing vehicle). It will be appreciated that the controller in this case includes a clock so that data gathered can be associated with a corresponding time.

[0073] In this case, the mine operates during the day, and the temperature sensor monitors the temperature rising and falling each day. Each sensor unit also detects the typical sound cycle of the day.

[0074] Regarding the image sensor, each sensor unit detects changes in the image throughout the day. It will be appreciated that each sensor unit will see a different image. However, the baseline reference frame can compare changes in the image as a function of time. So, for example, the baseline reference frame may include details on changes in the light intensity which will follow a daily cycle.

[0075] The passage of vehicles within a particular portion of the cycle (e.g., during the day) may be identified as part of the baseline reference frame during machine learning. For example, at the gate, any car (regardless of colour) may form part of the refence frame. However, on the road to the mine itself, where traffic is usually limited to trucks, the baseline reference frame may identify the movement of those trucks between the mine and the crusher to be part of the reference frame. A different vehicle on this road (e.g., different size or colour) would then be identified as a deviation or event.

[0076] For all sensor units, the baseline reference frame may be configured to identify that some short-term image changes may happen within particular time periods. For example, the passage of vehicles during operation hours may be categorised as part of the baseline reference frame, whereas the passage of vehicles outside operation hours may be categorised as a deviation or event. [0077] For example, the sensor unit beside the crusher may detect a significant increase in sound starting at 8am and continuing throughout the day until 6pm. The third and fourth sensor units may detect periodic sound events between 8am and 6pm associated with trucks passing between the quarry and the crusher. And the first sensor unit may detect periodic sound events associated with workers arriving just before 8am and leaving just after 6pm, as well as vehicular traffic throughout the working day.

[0078] It will be appreciated that each sensor unit may take several cycles to determine the baseline reference frame for each sensor unit. It will be appreciated that there may be several characteristic timescales that may be useful to observe. For example, in this example, we are focusing on the daily cycle. In other examples, the weekly cycle may also be important.

[0079] In this case, in response to detecting a sound event from the microphone sensor in one of the sensor units (e.g., sensor unit 200c), the sensor unit determines if there is a current wireless network available associated with the non-unique identifier for the sensor type (e.g., AUD for an audio sensor) with a matching unique identifier (e.g., an alphanumeric string such as 123456789ABCDEF). On detecting no networks matching the unique identifier and non unique sensor identifier, the sensor unit initiates the creation of a network called AUD_123456789ABCDEF to form a local network 212. In response to detecting an event, the sensor unit in this embodiment also increases the frequency of monitoring the environment via its own sensor array.

[0080] All sensor units 200a-d in the area periodically monitor for new wireless networks. On the detection of the wireless network initiated by sensor unit 200c, with the AUD code and the matching unique identifier, all sensors in communication range of the initiating sensor unit with a sensor matching the AUD type code (e.g., having an audio sensor such as a microphone) begin to increase the frequency of monitoring the environment using these matching sensors (e.g., audio sensors in this case) and to record detailed AUD data in a time limited local buffer for a predetermined time threshold after detecting the presence of the AUD network. The predetermined time threshold may be based on the type of event that is being detected. E.g., to monitor car traffic, the predetermined time threshold may be 5 minutes. If no events have occurred before this predetermined threshold, the initiated network may automatically expire. [0081] The other sensor units in the area (i.e., without audio sensors) do not join the AUD network because no matching sensor events have been detected. It will be appreciated that, in some embodiments, the sensor units may monitor the environment before the network is initiated, but at a lower frequency or intensity. The higher frequency monitoring may correspond to an active mode, and the lower frequency monitoring may correspond to a passive more.

[0082] On the detection of an audio event by a second sensor unit while the AUD_123456789ABCDEF network is live (e.g., within the predetermined time threshold), the sensor unit joins the AUD_123456789ABCDEF network and transmits all local buffer information associated with the audio sensor module of the second sensor unit to the root node with timestamps and sensor location. In this way, a dedicated ad hoc sensor unit network can be established in order to collate data on a particular event relating to a particular sensed parameter (e.g., in this case, audio information).

[0083] The geographical proximity may be determined, as in this embodiment, by each sensor unit having a geographical positioning sensor configured to determine a location for the sensor unit (e.g., via GPS), or by only connecting to units within the geographical range of the transceiver.

[0084] In some embodiments, the sensor units are configured to enable the creation of a local network such that all the sensor units in the local network are within a predetermined geographical range (e.g., all sensor units fit within a circle of radius 1 km). Other embodiments, the sensor units are configured to enable the creation of a local network such that each sensor unit within at most a particular distance from another sensor unit in the network. This may allow sensor units to pass information throughout the network, regardless of whether a sensor unit at one periphery can directly exchange information with another sensor unit at the opposite periphery.

[0085] As described above, once the baseline reference frame is determined for each sensor unit, each controller is configured to identify deviations from the baseline reference frame as an event, and to transmit data associated with the event via the transceiver.

[0086] Deviations can include a change in the value of a sensed parameter (e.g., a decrease in temperature or an increase in volume). This is particularly relevant for sensors configured to measure a single value. For sensors configured to monitor a spectrum (e.g., of light or sound), a deviation may include one of more of: changes in amplitude of the spectrum (as a whole or parts) and/or shifts of characteristic peaks to different frequencies. For sensors configured to record spatial information (e.g., an image sensor, or a height sensor) deviations may include detecting movement.

[0087] All deviations are determined with respect to the baseline reference frame which, in this example, reflects expectation values as a function of time determined by the sensor unit. Because, in this embodiment, the baseline reference frame includes temporal information, it will be appreciated that the same circumstances may or may not be determined to be an event depending on when they occur. For example, a vehicle passing the image sensor during the day may be expected, whereas the same vehicle passing the image sensor at night may be deemed to be an event.

[0088] During operation, each network root node communicates to a data storage server 211 (e.g., directly or via other sensor units in the network) and can periodically or continuously upload information as required. This information will include the baseline reference frame for each sensor unit (which can be used to predict events as a function of time), and the current data being reported in real time or after a certain time period as configured. In some embodiments, to minimize bandwidth, the baseline reference frame data is updated by transmitting changes to the baseline reference frame rather than retransmitting the baseline reference frame data again.

[0089] While monitoring the environment, each sensor unit communicates with additional sensor units in the area within a network. This may be done in different ways but, in this embodiment, each sensor uses a low power wireless mesh network to communicate with adjacent sensors in the immediate vicinity. During this communication phase, each sensor exchanges its shared baseline reference frame to adjacent sensors.

[0090] Sensor units which have a baseline reference frame sufficiently similar to received shared baseline reference frames of other sensor units may join with those other sensor units to form a local network. This local network group can communicate directly over a wireless mesh network and/or using communication tunnels over traditional backhaul networks such as cellular, satellite, or Wi-Fi. In the case of using communication tunnels, a network server must be available that lists a common ledger of all current IP addresses of sensors on the network.

[0091] Once a sensor unit has a baseline reference frame that accurately predicts the environment of the sensor unit this baseline reference frame is shared with a network coordinator. The network coordinator may be one of the sensor units, a computer connected to the local network, or a remote computer. The network coordinator shares baseline reference frames with all known sensor units in the local network. The shared baseline reference frames each include network identification and location information for the sensor unit that generated the baseline reference frame. The system may be configured to perform a test to ensure that the baseline reference frames are compatible to reduce sensor frame traffic between sensor units. This may include a regression test or machine learning with a threshold for similarity either determined through regression or as an output of the machine learning inference. Sensor units test the shared baseline reference frames and create mesh networks with sensor units that have compatible baseline reference frames by accessing the network coordinates. Networks can be formed in close proximity using mesh network technologies. Networks can be extended by tunneling through cellular or satellite networks.

[0092] From time to time the reference group can share updated shared baseline reference frames to collectively tune their own baseline reference frames to improve prediction. The reference local network will also become the source of network-wide events.

[0093] When the four sensor units 200a-d form a network 212 as shown in figure 2, they may share a baseline reference frame to predict temperature, noise background, and image background data. The system may be configured to share a baseline reference frame based on a measure of stability. For example, components of the sensor inputs may be compatible while others are not because some measured parameters may be consistent across the sensor network while others are not. For example, the noise sensor baseline reference frames may be incompatible because noise can vary significantly over short distances. In contrast, temperature, humidity, wind speed and direction changes, vibration, may more compatible as they would be more consistent across the area of the local network. The system may build a baseline reference frame based on particular sensed parameters, each sensor having a baseline reference frame. This would allow a multi-sensor baseline reference frame to be created using multiple sensed parameters to create a more detailed model of the environment.

[0094] When an event happens, the event is shared on a temporary network with a network ID, the root node may form an event chain. When a vehicle drives by the first sensor unit out of hours, this is identified as an event because it represents a significant deviation from the baseline reference frame (in terms of image and sound). The first sensor unit captures image (e.g., OPT for optical) and sound (AUD) data.

[0095] A network is created using the ALIDOPT tag and sensor units in range increase their sensor polling frequency and storing temporary time limited local buffers of the specified sensors. When the additional sensors detect an event (e.g., sound and/or visual sensors), they join the initiated or established network to share event information. An event notification is sent with a time stamp, event data, location and an event identifier to each of the sensor units in the networked group. The time stamp may be a point in time (e.g., 12:14 on 18/08/2022) or a time period (e.g., 12:14-12:18 on 18/08/2022). In this case, the other sensor units are configured to monitor for events within a predetermined time period (e.g., 10 minutes) of the received time stamp and/or with similar deviation characteristics (e.g., a vehicle of similar size and colour and/or with a similar engine noise). [0096] In this case, the vehicle drives to the crusher and the sound is detected by the second sensor unit 200b. The second sensor unit joins the ALIDOPT network and sends temporary buffer data leading up to the event, and the event notification which includes time stamp information and location information. This information is associated with the event identifier received from the first sensor unit on the root node. The second sensor unit may update the timestamp of the event (e.g., by moving the point in time or by extending the extent of the time stamp period). The root node stores the event information from the second sensor unit to the reference group event chain with this new data.

[0097] If the vehicle passed the third and fourth sensor unit, the event data could be updated in the same way, and all the data associated with that event would be associated with the same event identifier or network ID. This means that the sensor units themselves can assemble an event comprising information derived from multiple sensor units within the ad hoc network. In this way, the sensor units themselves are performing the process of associating disparate pieces of information corresponding to an event. This information can then be transmitted to a central server for access by a user.

[0098] That is, instead of a series of isolated events which would need to be analyzed remotely, the system would connect multiple events into an event chain that can be used to derive additional insights from the data from the sensors in the field, without requiring data analytics in the cloud to create these data insights. [0099] For sensor units that can be in the area but out of range of the wireless network used to locally connect sensors, a cloud system may be used to coordinate event notifications associated with the unique asset identifier and the sensors used in the event. In this way, virtual wide area event networks are created using event clustering. Each root node holding the event notification data will be given the remaining cluster information through data download from the cloud architecture so that each root node has a full event picture that is used for local data interpretation.

[0100] It will be appreciated that proximity in time, and similarity in sensor data allows the sensor units to associate different sensor readings with the same event. For more complex situations, the system may be configured to use these temporal and sensor-data characteristics to identify overall events which share the same general characteristics and sub-events which share some, but not all, of the same general characteristic. These subevents may be collated into sub-event chains, each sub-event chain comprising information from multiple sensor units relating to a particular aspect of an associated event chain.

[0101] For example, if there are multiple vehicles arriving unexpectedly at the site at around the same time, the system may be record this as an event chain, but to track the movement of the individual vehicles around the site as separate sub-events. The tracking may be based on, for example, the characteristic shape, colour or even licence plate of each vehicle (based on image sensor data) and/or the characteristic sound of each vehicle. In another example, the sensor units may be configured to identify the repeated presence of a particular vehicle as a single event chain, and each instance of its presence as a sub-event. In this case, the timing of each deviation would be far apart, but the sensor characteristics of each sub-event would be similar.

[0102] Similarly, with dust, wind direction, gas leaks, etc. it should be possible to create event chains related to a source location, even when there are several locations.

Crop Monitoring

[0103] Figure 3 shows a further embodiment which is two fields beside each other, each with six sensor towers arranged around the periphery. In this case, both fields have crops: a barley field and a potato field. [0104] In this case, the farmer set up the sensor units when he first ploughed the field. This allows the sensor units to begin to gather data and monitor the fields throughout the growing season.

[0105] In this embodiment, each sensor unit comprises: a humidity sensor, a spectral analyser to determine the colour of light, a height sensor which can measure the height of the plants as they grow throughout the season, and a temperature sensor.

[0106] As with the sensor units of figure 1 , each sensor unit in this case has a transceiver configured to wirelessly receive and transmit data with other sensors in a local network; and a controller configured to receive and process data from the sensor array, and to receive, transmit and process data via the transceiver.

[0107] Like the previous system of figure 2, each sensor unit in this system is configured to determine a baseline reference frame, the baseline reference frame providing a baseline in cyclic changes in the sensor data. This baseline reference frame includes the colour and height of the canopy below the sensor unit (e.g., to the earth or the top of the plants in the vicinity of the sensor), and how the humidity and temperature changes over time.

[0108] In this embodiment, each sensor unit is configured to form a local network with multiple other sensor units via the transceiver based on geographical proximity and a degree of similarity of the determined baseline reference frames of the sensor unit and baseline reference frames of the multiple other sensor units. In this case, the daily cycle of the two fields is initially similar. Therefore, the sensor units initially form a local network which includes all twelve sensor units.

[0109] However, when the barley starts to sprout before the potatoes, the dominant spectral colour of the six sensor units in the barley field register this as an event. A corresponding event is not yet determined for the potato field.

[0110] Unlike the detection of a vehicle as described in relation to figure 2, the situation brought about by the sprouting event persists, and the six barley field sensors continue to observe a dominant green spectrum for several cycles. In this embodiment, after several cycles, the six barley field sensor units and the six potato field sensor units identify that their baseline reference frames are incompatible. However, the six barley field sensor units also recognise that they are compatible with each other, and separately, the six potato field sensor units recognise that they are compatible with each other. In response to determining which sensors are sufficiently similar to each other, they divide the sensor units into two local networks. In this way, the sensor units themselves form an appropriate local network based on similarity of the shared baseline reference frames.

[0111] It will be appreciated that the event of the barley sprouting may be identified as an event first by one of the sensor units in the barley field. This may cause the first sensor to initiate a local network with a network ID comprising a non-unique component identifying the image sensor (e.g., OPT) and a unique component. When the other sensors in the field detect the barley sprouting using their image sensors, they may then look for a live network with a network ID identifying the image sensor type, and join that network to exchange data on the event. In this example, the first sensor unit to detect the event may be designated as the root node in the local network. It will be appreciated that the event in this example occurs over a much longer timescale than the vehicle event in the last example. Therefore, the time period over which the local network is live is longer (e.g., a week). Within this time period, the sensor units may be in an active mode and determining the colour of the barley field more frequently.

[0112] As before, once the baseline reference frame is determined, the controller is configured to identify deviations from the baseline reference frame as an event, and to transmit data associated with the event via the transceiver.

[0113] Over the season the plants grow. The baseline reference frame is updated as the year progresses as the deviation in height of the plants is gradual enough not to trigger an event.

[0114] Later on in the season, the barley ripens, and the potatoes start to die off. For the barley field, this is associated with a change in colour from green to yellow. In a field, this does not necessarily happen evenly, and often parts of a field will ripen slightly before the rest (e.g., due to variations in moisture and soil chemistry). When the barley starts to ripen (e.g., in the north-west corner), that sensor unit will register the change in colour spectrum as an event. Over the next few days corresponding events may occur for the other sensor units as the entire field ripens. In this embodiment, the sensor units are configured to associate events identified with different sensor units over a seven-day period with a single event identifier (or network ID) if the events identified with different sensor units have corresponding features (e.g., a move from a green spectrum to a yellow spectrum). Again, the first sensor unit to detect the ripening event may be designated as the root node, and information from the other sensor units relating to this event may be passed to the root node.

[0115] Later the potato tops (e.g., stems and leaves) die off, and this is associated with a change in colour (from green to brown) and a decrease in the height of the plants which can be detected by the image and height sensors and event. Like the ripening of the barley, this can vary across the field. When the potato tops start to die off (e.g., in the middle), those sensor units will register the change in colour spectrum as an event. Over the next few days corresponding events may occur for the other sensor units as the entire field dies off. In this embodiment, the sensor units are configured to associate events identified with different sensor units over a seven-day period with a single event identifier (or network ID) if the events identified with different sensor units have corresponding features (e.g., a move from a green spectrum to a brown spectrum).

[0116] It will be appreciated that the system would also be able to determine evidence of disease or issues (e.g., premature dying due to drought or disease) and/or to monitor the progress of a particular treatment (e.g., the reduction of a particular spectral line associated with a weed leaf colour following a herbicidal spray treatment).

[0117] In this way, the farmer is able to easily set up sensor units in fields in such a way that they themselves can organise the appropriate size and extent of the local network.

Initiating and Establishing a Local Network

[0118] Figure 4 is a flow chart showing how the sensor unit interacts with other sensor units within range to establish ad-hoc networks for communicating data relating to an event or event chain.

[0119] Each sensor unit is configured to periodically monitor to determine whether another local sensor unit in the area has initiated or established a live local network. In this example, a sensor unit will initiate a local network in response to detecting an event using one or more sensors in that sensor unit’s sensor array. The local network will have a network ID which identifies the sensor type used to detect the event.

[0120] In response to a first sensor unit detecting a local network initiated by a second sensor unit in the area, the first sensor unit increases the polling frequency of the sensors in the first sensor unit’s sensor array having the same sensor type as those identified in the network ID. Increasing the polling frequency of the sensors may comprise monitoring the environment using that sensor more frequently and/or storing more sensor data in a buffer. This may be considered an active mode. If the first sensor unit does not detect that a local network has been initiated, the first sensor maintains the polling frequency of the sensors. This may be considered a passive or power-saving mode.

[0121] Whether in a passive mode or in active mode, the first sensor in this example is also configured to identify an event by identifying deviations from an established baseline for each of the sensors in the sensor array. In response to detecting an event, the first sensor unit is configured to determine whether the event has been detected by a sensor type corresponding to a detected network ID. For example, if the first sensor detects an audio event, the first sensor may determine whether any detected network IDs include the non-unique identifier “AUD”.

[0122] If a detected has been initiated or established, the first sensor unit will join that local network and transmit sensor data associated with the event detected by the first sensor unit to that network.

[0123] If no network associated with the sensor type which has detected the event data on the first sensor unit, the first sensor unit will initiate a local network and create a unique network ID comprising a non-unique portion identifying the sensor type or types used to identify the event, and a unique portion uniquely identifying the local network.

[0124] In this example, a live local network (e.g., an initiated or established local network) may be configured to expire following a predetermined period of time following initiation or following the last data associated with an event associated with the network ID. It will be appreciated that once a local network has expired, data is no longer exchanged using that network ID.

[0125] In relation to the structure of the local network, typically the sensor unit which initiates the local network is the root node. Secondary sensor units which detect subsequent deviations in their own sensor data and connects to the initiated local network using the associated network ID, are child nodes and transfer their sensor data to the parent node.

[0126] In response to a sensor unit detecting an event with more event sensor types than are identified in the network ID associated with a said live temporary local network (e.g., but there is at least one sensor type in common), the sensor unit creates a new network as a root node and all event data is transferred to the new root node. [0127] For example, if a first sensor detects an audio event, the first sensor may initiate a local network with a network ID comprising a non-unique component identifying the sensor type as audio (e.g., AUD). A second sensor may then detect an event comprising sound (AUD) and image (OPT) information. In this case, rather than joining the network initiated by the first sensor unit, the second sensor unit recognises that although the event that it has detected has a common sensor type with the initiated local network (AUD in this case), it has additional network types (OPT in this case) and so initiates and establishes a new local network with the first sensor unit. The new local network has a network ID with both AUD and OPT identifiers. Information is then passed from the first sensor unit to the second sensor unit and the second sensor unit is identified as the root node. The unique identifier for the new audiovisual local network may or may not be the same as that for the initial sound only local network. This allows the root node to have the capacity to store data from all the sensor types identified in the event.

Other Options

[0128] If a sensor is installed adjacent to another sensor, it may be that the other sensor is indoors, or in a vehicle, etc. These sensors would have completely different baseline reference frames, and so would not join for form a network. This illustrates that using the geographical proximity of the sensors alone may not be sufficient to determine the most appropriate extent of the local network.

[0129] All sensors also have a geographical “closeness” that can be used to separate sensors from forming reference groups. This could also be defined by a user who owns a fleet of sensors. Sensors outside of the fleet do not need to talk to other sensors (e.g., based on data identifying the owner) but could be configured to do so.

[0130] The present technology may use a hybrid network architecture hardware. A hybrid network is any network that uses more than one type of connecting technology or topology (e.g., cellular and Wi-Fi). This hybrid network architecture helps allow groups of devices to communicate and form mesh networks without the need for any dedicated equipment infrastructure. The communication is enhanced by using available infrastructure such as cellular or satellite to tunnel information between networks thereby allowing for large disparate networks to be remotely connected.

[0131] The present technology may use this multiprotocol communication layer to transfer information to other networks about the network baseline reference frame. The baseline reference frame includes a prediction of the network state for the next block of time. E.g., a prediction based on a set of sensor data that represents what is going to happen over a block of time. Like networks that have similar baseline reference frames group together to create baseline reference frame based virtual networks.

[0132] At the end of a baseline reference frame time interval or cycle (e.g., a day), insights are exchanged by members of the baseline reference frame based virtual networks. If consensus is reached on coincident insights, the information is added to the baseline reference frame and the baseline reference frame is updated across the baseline reference frame based virtual network and it becomes the new baseline reference frame. The change may be stored as a network wide event.

[0133] If insights are isolated in one baseline reference frame period, the insights become a local event and are stored. If the local events occur in multiple baseline reference frame periods, the network connected sensor unit leaves the baseline reference frame based virtual network.

[0134] All baseline reference frame virtual networks process information at the edge collaboratively and without ever needing a centralized service. Data can be stored and archived on the network connected sensor and retrieved via local exchange (e.g., a drone) or transferred to any network endpoint for archiving.

[0135] If a new sensor unit is brought into a region which already has an established local network, the new sensor unit may be configured to initiate communication with other sensor units in the area. Then, as described above, the new sensor unit can become part of the established local network based on whether or not the determined baseline reference frames are sufficiently similar. This may allow a new sensor unit to be added to a local network after a local network has been established.

[0136] When a new sensor unit communicates with sensor units in an established local network, the local network may respond by providing one or more baseline reference frames to the new sensor unit. This may allow the new sensor unit to determine its baseline reference frame more quickly as it has a baseline to start with, to more quickly identify events (as opposed to features of the baseline reference frame) and/to determine whether its baseline reference frame is compatible with the established local network.

[0137] The assembly may be configured to determine an event reference frame, the event reference frame being a collection of consistent deviations of sensor data recorded by multiple sensor units across multiple identified events. For example, a series of sensor units may be set up on a road and form a local network together. If a car passes one sensor unit, it will typically pass another sensor unit at a certain time depending on the speed of the car. The system can then know, based on a car passing one sensor unit and on previous car-passing events, when the car will be detected on other sensor units within the sensor network assembly. The assembly may therefore be able to detect an event deviation and an associated abnormal event based on deviations from the event reference frame. For example, if the car arrives at another sensor unit before its predicted time (e.g., speeding) or if it does not arrive at its predicted time (e.g., the car has stopped or went off the road). This event reference frame is based on data from multiple sensor units within the local network.

[0138] Although the present invention has been described and illustrated with respect to preferred embodiments and preferred uses thereof, it is not to be so limited since modifications and changes can be made therein which are within the full, intended scope of the invention as understood by those skilled in the art.

Claims

1. A sensor unit comprising: a sensor array of one or more sensors for monitoring the environment around the sensor unit; a transceiver configured to wirelessly exchange data with other sensors via a temporary local network, the temporary local network being associated with a network ID, the network ID comprising a unique identifier and information identifying at least one sensor type; and a controller configured to: process data and to exchange data with the sensor array and with the transceiver, the controller being configured to determine a baseline reference frame using data received from the sensor array, and, to identify deviations from the determined baseline reference frame as an event; and in response to receiving a said network ID from another sensor unit, increase a polling frequency of the sensors in the sensor array of the at least one sensor type identified of the received network ID; and wherein, in response to detecting an event using one or more of the sensors of the sensor array, the sensor unit is configured to exchange sensor data associated with the detected event using a said temporary local network associated with a said network ID, wherein, where the sensor unit has identified that a temporary local network has been initiated with a said network ID identifying a sensor type corresponding to the at least one sensor used to identify the event, the sensor unit is configured to transmit the sensor data via the identified temporary local network, and wherein, where the sensor unit has determined that no local temporary network has been initiated with a network ID identifying a sensor type corresponding to the at least one sensor used to identify the event, the sensor unit is configured to initiate a temporary local network by creating a network ID and enabling transmission of the created network ID to other sensor units.

2. The sensor unit according to claim 1 , wherein, in response to a sensor unit detecting an event with more event sensor types than are identified by a said network ID associated with a said first live temporary local network, the sensor unit is configured to create a new second live temporary local network in which the sensor unit is a root node, and to allow all event data from the first live temporary local network to be transferred to the new second live temporary local network.

3. The sensor unit according to any one of claims 1-2, wherein the sensor unit is configured to communicate with a local group of wirelessly connected sensor units via the transceiver, wherein the local group forms a wireless mesh or star network.

4. The sensor unit according to any one of claims 1-3, wherein the sensor unit is configured to periodically exchange baseline reference frame data with other sensor units of a local network, and to refine the baseline reference frame using the received baseline reference frame data.

5. The sensor unit according to any one of claims 1-4, wherein the sensor unit is configured to associate deviations from the baseline reference frame with a unique event identifier and a timestamp.

6. The sensor unit according to claim 5, wherein the controller is configured, in response to receiving event data comprising a timestamp and a unique event identifier from another sensor unit, the received timestamp being within a predetermined period before an identified deviation detected by the sensor array, to associate the identified deviation with the received unique event identifier, so that the unique event identifier is associated with an event chain comprising event data from multiple sensor units.

7. The sensor unit according to claim 5 or claim 6, wherein the controller is configured, in response to receiving event data comprising sensor data and a unique event identifier from another sensor unit, the received sensor data being sufficiently similar to sensor data associated with an identified deviation detected by the sensor array, to associate the identified deviation with the received unique event identifier, so that the unique event identifier is associated with an event chain comprising information from multiple sensor units.

8. The sensor unit according to any one of claims 1-7, wherein the controller is configured to associate two events detected at different sensor units when the events have sufficiently similar sensor signatures.

9. The sensor unit according to any one of claims 1-8, wherein the sensor unit is configured, when the sensor unit identifies a said temporary local network after detecting an event, to join the identified temporary local network as a child node.

10. The sensor unit according to any one of claims 1-9, wherein the sensor unit is configured, when the sensor unit initiates a local network after detecting an event, to designate the sensor unit as a root node of the initiated local network.

11. The sensor unit according to any one of claims 1-10, wherein the sensor array comprises one of more of: a temperature sensor; a humidity sensor; and an ambient light sensor.

12. The sensor unit according to any one of claims 1-11 , wherein the sensor array comprises one or more of: wind speed sensor; a wind direction sensor; a gas detector; and a height sensor.

13. The sensor unit according to any one of claims 1-12, wherein the sensor unit is configured to decide whether or not to separate from an established local network connection with another sensor unit based on the degree of similarity between the baseline reference frames of the sensor unit and the other sensor unit.

14. The sensor unit according to any one of claims 1-13, wherein the sensor unit is configured to initiate communications with other sensor units in the vicinity in response to determining that the position of the sensor unit has changed.

15. The sensor unit according to any one of claims 1-14, wherein the sensor unit comprises multiple transceivers configured to receive and transmit data using multiple different communication protocols.

16. The sensor unit according to any one of claims 1-15, wherein the local network is a hybrid network in which sensor units are configured to communicate using multiple different network protocols.

17. The sensor unit according to any one of claims 1-16, wherein the controller is be configured to change the mode of the sensor unit from a passive mode to an active mode in response to receiving event data from another sensor unit of the local network.

18. The sensor unit according to any one of claims 1-17, wherein the sensor unit has one or more sensors for measuring the same environmental parameter.

19. The sensor unit according to any one of claims 1-18, wherein the sensor unit is capable of joining a wide area network to receive event information from other sensor units which are outside the range of the transceiver.

20. A method of forming a local network between a plurality of sensor units, each sensor unit comprising: a sensor array of one or more sensors for monitoring the environment around the sensor tower; a transceiver configured to wirelessly receive and transmit data with other sensors of a local network; and a controller configured to receive and process data from the sensor array, and to receive, transmit and process data via the transceiver; wherein the method comprises each sensor unit: determining a baseline reference frame, the baseline reference frame being a baseline of sensor array data over the course of a cycle period, the baseline reference frame being determined using data received from the sensor array over multiple cycles while the sensor unit is stationary; once the baseline reference frame is determined, identifying deviations from the baseline reference frame as an event, and to transmit data associated with the event via the transceiver; and in response to receiving a network ID, increasing the polling frequency of the sensors of the sensor array of the at least one sensor type identified by the network ID; in response to detecting an event using one or more of the sensors of the sensor array, transmitting sensor data associated with the detected event using a temporary local network associated with a said network ID, wherein, where a live temporary local network has been initiated with a network ID identifying a sensor type corresponding to the at least one sensor used to identify the event, the sensor data is transmitted via the live temporary local network, and wherein, where no live local temporary network has been initiated with a network ID identifying a sensor type corresponding to the at least one sensor used to identify the event, initiating a temporary local network by creating a network ID and transmitting the created network ID to other sensor units.