Adaptive radio configuration in wireless networks

Abstract

To provide adaptive radio configuration in wireless networks.SOLUTION: A wireless networking system is provided. The wireless networking system includes a base station device including processing circuitry configured to detect a transmission rate from a portion of a preamble of an incoming packet transmission signal and adapt a radio configuration to receive a remainder of the incoming packet transmission signal at the transmission rate.SELECTED DRAWING: Figure 1

Description

[Background technology]

[0001] Low-power, long-range wireless networks such as LoRa (Long Range) are becoming increasingly mainstream for Internet of Things deployments. Given the versatility of applications these protocols enable, they support many data rates and bandwidths. However, for a given network deployment, which may span several miles, network operators must specify the same configuration, or a small subset of configurations, for all devices in the network to communicate with each other. This one-size-fits-all approach is highly inefficient for large networks that may span several miles and have hundreds of devices, as many, if not most, of the wireless devices connected to the low-power, long-range network's base station (gateway) often experience less-than-optimal data transmission rates. Summary of the Invention

[0002] A wireless network system is provided that includes a base station device that includes processing circuitry configured to detect a transmission rate from a portion of a preamble of an incoming packet transmission signal and adapt a radio configuration to receive the remainder of the incoming packet transmission signal at that transmission rate.

[0003] This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. Moreover, the claimed subject matter is not limited to implementations that solve some or all of the disadvantages noted in any part of this disclosure. [Brief explanation of the drawings]

[0004] [Figure 1] 1 shows a schematic diagram of a wireless network system according to an embodiment of the present disclosure. [Figure 2] 2 shows a schematic diagram of a transmission packet analyzed by the wireless network system of FIG. 1; [Figure 3] 3 shows a schematic diagram of a base station device reading the transmission packet of FIG. 2; [Figure 4A] 2 shows a graph illustrating data rates and preamble structures for transmission packets such as those of FIG. 1; [Figure 4B] 2 shows a graph illustrating data rates and preamble structures for transmission packets such as those of FIG. 1; [Figure 5] A schematic diagram of a base station device such as that shown in Figure 3 is configured with a software-designed radio. [Figure 6A] 3 shows a graph illustrating a sampling method for sampling transmission packets such as those of FIG. 2; [Figure 6B] 3 shows a graph illustrating a sampling method for sampling transmission packets such as those of FIG. 2; [Figure 6C] 3 shows a graph illustrating a sampling method for sampling transmission packets such as those of FIG. 2; [Figure 7] 3 shows a graph illustrating adaptive sampling of a transmission packet such as that of FIG. 2. [Figure 8A] Figure 2 shows a spectrum diagram of the LoRa upchirp included in the transmission packet. [Figure 8B] Figure 2 shows a spectrum diagram of the LoRa upchirp included in the transmission packet. [Figure 8C] Figure 2 shows a spectrum diagram of the LoRa upchirp included in the transmission packet. [Figure 9A] 2 shows a graph illustrating data characteristics of transmission packets used by the wireless network system of FIG. 1; [Figure 9B] 2 shows a graph illustrating data characteristics of transmission packets used by the wireless network system of FIG. 1; [Figure 9C]2 shows a graph illustrating data characteristics of transmission packets used by the wireless network system of FIG. 1; [Figure 10] 4 shows a coverage map illustrating the supported combinations of spreading factors and bandwidths supported by the base station device of FIG. 3; [Figure 11] 2 shows a schematic diagram of a multi-stage artificial intelligence model used in the wireless network system of FIG. 1; [Figure 12A] 12 shows a schematic diagram of the neural network used in the multi-stage artificial intelligence model of FIG. 11. [Figure 12B] 12 shows a schematic diagram of the neural network used in the multi-stage artificial intelligence model of FIG. 11. [Figure 13] 2 shows a graph illustrating the accuracy of the wireless network system of FIG. 1 compared to other systems. [Figure 14A] 2 shows a graph illustrating the accuracy of the wireless network system of FIG. 1 across various bandwidths, spreading factors, and locations. [Figure 14B] 2 shows a graph illustrating the accuracy of the wireless network system of FIG. 1 across various bandwidths, spreading factors, and locations. [Figure 14C] 2 shows a graph illustrating the accuracy of the wireless network system of FIG. 1 across various bandwidths, spreading factors, and locations. [Figure 14D] 2 shows a graph illustrating the accuracy of the wireless network system of FIG. 1 across various bandwidths, spreading factors, and locations. [Figure 15A] 2 shows a graph illustrating the accuracy of the wireless network system of FIG. 1 over various locations and times. [Figure 15B] 2 shows a graph illustrating the accuracy of the wireless network system of FIG. 1 over various locations and times. [Figure 15C] 2 shows a graph illustrating the accuracy of the wireless network system of FIG. 1 over various locations and times. [Figure 16A] 1 shows a flowchart of a method according to an embodiment of the present disclosure. [Figure 16B]1 shows a flowchart of a method according to an embodiment of the present disclosure. [Figure 17] 2 illustrates an exemplary computing environment in which the system of FIG. 1 may be implemented. DETAILED DESCRIPTION OF THE INVENTION

[0005] To address the above issues, FIG. 1 illustrates an exemplary wireless network system 100 configured to allow network devices to transmit at any data rate. The wireless network system 100 includes a base station device 102 that uses the first few symbols of a preamble 107A1 of a packet transmission signal 107A to classify the correct data rate, switch the base station's radio configuration, and then decode the data. The design of the present disclosure exploits the inherent asymmetry in outdoor IoT deployments where clients are power-hungry and resource-constrained, but the base station device 102 (i.e., wireless gateway) is not (the terms base station and wireless gateway are used interchangeably herein). The wireless network system 100 disclosed herein is backward-compatible with the existing LoRa protocol and accurately identifies the correct configuration with over 97% accuracy in both indoor and outdoor deployments.

[0006] Section 1: Introduction Low Power Wide Area Networks (LPWANs) like LoRaWAN are becoming increasingly popular for large-scale Internet of Things (IoT) deployments. Despite being early days, there are already over 100 million devices using LoRaWAN in deployments, and this number is expected to exceed 730 million by 2023. LPWANs can operate at low power, communicate over long distances, and are low cost compared to other mainstream solutions. These characteristics make such wireless-based devices ideal for low-throughput, large-scale networks in cities, agriculture, forestry, and many other industries.

[0007] To support long distances and diverse device requirements, LoRaWAN can operate at many different data rates. Data rates are configured using two parameters: the bandwidth (BW) and spreading factor (SF) of the chirps used in LoRa transmissions, as shown in Figure 4A. Figure 4B shows a spectrum diagram of a LoRa preamble with eight up-chirps and two down-chirps. The actual data rate also depends on the code rate used to ensure error correction. A fixed code rate is assumed. As expected, higher bandwidths allow for higher data rates. The spreading factor defines the time it takes to transmit one chirp; a higher spreading factor means a longer time to transmit the signal and therefore a lower data rate. A typical LoRa implementation can support bandwidths from 7.8 kHz to 500 kHz and spreading factors (log scale) from 7 to 12. Thus, a device transmitting at 7.8 kHz and a spreading factor of 12 achieves a data rate that is approximately 1189 times lower than a device transmitting at 500 kHz and a spreading factor of 7.

[0008] Despite this wide range of possibilities for devices in a network, the current paradigm requires system designers to configure a single configuration setting (or a small subset of compatible configuration settings) for the bandwidth and spreading factor of the entire network. That is, the bandwidth and spreading factor are the same for all devices. LoRaWAN Automatic Data Rate (ADR) algorithms have been proposed, but they can take hours or days to converge, have significant control overhead, and do not address multiple bandwidths. As a result, for example, in a farm network, the network is typically configured to connect to the furthest-located tractor-like device on the farm, even though most networked sensors or even tractors on the farm are usually located near the network's nearest wireless base station device. This design choice arises from the need to limit complexity in the network and reduce the control overhead of coordinating frequent data rate changes. However, this design choice has three significant drawbacks, described below.

[0009] Network Throughput LPWAN devices operate over large areas. A single LoRa gateway (LoRaWAN uses a gateway-client mode of operation) is designed to cover a range of approximately 10 km with up to several thousand devices. In such large-scale deployments, devices at the end of the range can barely support lower data rates. As a result, this "single-size-fits-all" design forces even devices that can support higher data rates to operate at extremely low data rates. This reduces overall network throughput and reduces the number of devices the network can support by up to two orders of magnitude.

[0010] Deployment overhead The optimal configuration of the gateway needs to be set by the network operator. Typically, this is achieved by testing multiple configurations and selecting the one that works for all client devices. This process requires technical effort and is not always available, for example, when deploying such devices in remote rural areas for agricultural monitoring. Second, the selection of the configuration needs to be dynamic: due to changes in the environment or gradual deployment of devices, this configuration may stop working for a subset of devices over time, requiring frequent updates.

[0011] Mobility IoT devices may be mounted on moving vehicles such as tractors, buses, or pickup trucks. The optimal configuration changes as the vehicle moves and is difficult to predict before the move. The lowest data rate configuration may be selected, but this may significantly reduce the capacity of the network.

[0012] Presented herein is a novel wireless network system 100 that can support wireless devices 101 transmitting at different data rates. Each wireless device 101 transmits at its best possible data rate, which may depend on signal quality and application requirements, without the wireless device 101 having to notify the base station device 102 in advance, i.e., before initiating wireless communication, about the configuration of the wireless device 101. The approach described herein does not require the wireless device 101 to transmit any control packets, does not require modifications to the LoRa protocol, and is backward compatible with existing devices (i.e., does not require any hardware modifications to IoT wireless devices 101 that use the LoRa protocol).

[0013] As shown in FIG. 5, the wireless network system 100 uses a software-defined radio (SDR) in front of the LoRa transceiver in the wireless base station 104, which functions as a wireless gateway. The SDR detects the preamble, determines the signal's bandwidth and spreading factor from the preamble, and adjusts (i.e., adapts) the radio configuration of the LoRa radio to the appropriate setting to receive the packet. This allows the wireless gateway to successfully receive packets from clients operating in either configuration. Because this approach operates on a packet-by-packet level, it supports changing data rates due to client movement and dynamic changes in the environment. Within the wireless network system 100, there is a set of neural networks that use a small number of samples from the LoRa chirp to classify the correct radio configuration at the base station for each incoming packet transmission signal 107A.

[0014] The disclosed wireless network system 100 configuration addresses the following three technical goals and the associated challenges to achieving these technical goals in practical deployments:

[0015] sensitivity To maintain the long-term aspects of LoRa deployment, a first potential technical goal for the wireless network system 100 is to be able to operate at low signal to noise ratios (SNR).

[0016] Real-time operation A second potential technical goal of the wireless network system 100 is to be able to reconfigure the LoRa radio in real time to have enough time to detect a packet via SDR and properly receive the remainder of the packet signal, thereby ensuring that no packets are lost.

[0017] Compatibility with existing deployments A third potential technical goal is for existing deployments to require no protocol or client hardware changes to existing LoRa devices, while future generations of devices will not be so limited by this constraint.

[0018] The challenges associated with meeting these technical goals are described below, and an overview of the system of the present disclosure in Section 2 follows the discussion of the challenges.

[0019] The wireless network system 100 of the present disclosure takes a new approach to the basic rate adaptation problem in mobile networks. It does not require the client device and gateway to agree on a rate in advance. One might instead borrow from existing rate adaptation protocols, such as Wi-Fi, where the lowest data rate is used to transmit a preamble containing the data rate configuration. Such an approach is not suitable for LPWANs, as LPWANs are primarily designed for large-scale deployments where each client needs to transmit a small amount of data. Furthermore, because data rate variation in LPWANs is higher than in Wi-Fi, this causes very high overhead for packets transmitted at high data rates with small amounts of data (one symbol at the lowest data rate is 1,189 times longer than one symbol at the highest data rate). Furthermore, this adds hardware complexity to the client device design and does not directly account for the different bandwidths used by clients in LoRa.

[0020] The gateway of the wireless network system 100 is implemented using a Universal Software Radio Peripheral (USRP) SDR platform, which in one implementation can be realized with an off-the-shelf LoRa chipset as the client. The wireless network system 100 has been evaluated in a wide range of settings, such as benchtop experiments with various signal strengths, indoor deployments across multiple rooms, and outdoor deployments. The results are summarized below.

[0021] In testing, the configuration detection algorithm of wireless network system 100 was able to detect the correct coding parameters of incoming packet transmission signals with accuracies of 99.8%, 95%, and 98.2% in indoor, outdoor, and benchtop experiments, respectively, compared to the autocorrelation baseline, which achieved accuracies of 67.4%, 67%, and 78%, respectively.

[0022] The wireless network system 100 continues to operate effectively at low SNRs, achieving 94% accuracy even when the signal is attenuated by over 140 dB.

[0023] The algorithms of the wireless network system 100 can generalize effectively to new environments and continue to work in dynamic environments over time. In experiments over a five-day period, the accuracy of the wireless network system 100 was consistently above 99%, even with slight day-to-day variations.

[0024] Finally, it is recognized that the wireless network system 100 disclosed herein is applicable to future generations of devices. As neural networks evolve and faster hardware implementations are developed, the algorithms described herein can be adapted to shift the burden of rate adaptation solely to the powered base station infrastructure (i.e., to the gateway / base station device 102, rather than requiring coordinated configuration of both the base station device 102 and the mobile wireless device 101, as is the current situation), thereby alleviating the rate configuration overhead from battery-powered mobile devices. Thus, the techniques described herein are not limited to application to low-power wide area networks, but are also potentially applicable to various other types of wireless networks, including high-speed networks such as so-called sixth-generation (6G) wireless networks currently under development.

[0025] Section 2: Assignments As noted above, the wireless network system 100 disclosed herein aims to achieve three objectives: sensitivity, real-time operation, and compatibility. However, each of these objectives is difficult to achieve alone.

[0026] Let's start with sensitivity. Sensitivity in LPWAN protocols is directly related to bandwidth. Low-bandwidth signals experience less noise and can be received with lower signal strength. Conversely, higher-bandwidth signals require higher signal strength at the receiver to be correctly decoded. Thus, if the wireless network system 100 configures its SDR to operate at a low bandwidth, it will meet the sensitivity requirements but miss signals received at higher bandwidths. On the other hand, if the wireless network system 100 sets its bandwidth too high, it may miss signals coming from longer distances at lower bandwidths (and therefore lower signal strengths).

[0027] Second, to ensure real-time operation, it is desirable for the SDR to identify the correct configuration of a packet using only a few symbols. However, the length of the symbols themselves depends on the configuration used by the transmitter. A symbol transmitted using a spreading factor of 12 will be 64 times longer than a symbol transmitted using a spreading factor of 6. If the signal is sampled for too long, there is a risk of missing an entire packet for the highest data rate transmitter. On the other hand, if the signal is sampled for too short a period, there may not be enough information to identify the correct coding parameter configuration for a lower data rate transmitter.

[0028] Finally, to ensure backward compatibility, it is desirable for the wireless network system 100 to receive the entire packet after the appropriate configuration has been set at the gateway. However, this requires that the configuration be identified even before the signal reaches the gateway, a seemingly impossible task. These challenges are visualized in FIG. 6, which illustrates the challenges associated with configuring the receiving SDR itself. The figure shows chirps for three different configurations that are relatively close to each other. It is recognized that significant differences exist, but due to the large scale of the differences, it is difficult to visually represent such significant differences in such a plot. As shown in FIG. 6A, sampling one symbol at maximum bandwidth captures only a small portion of the high-bandwidth signal, reducing sensitivity. On the other hand, if one symbol length is sampled at the low-data-rate configuration in FIG. 6B, high sensitivity is maintained but significant delay is introduced for the high-data-rate symbol(s). Finally, one may wonder why the minimum values ​​for both frequency bandwidth and symbol duration are not used across all possible configurations. This ensures both sensitivity to low signal strengths and real-time operation. However, as shown in FIG. 6C, such a configuration would miss some configurations altogether.

[0029] To resolve this conflict between sensitivity and real-time operation, the wireless network system 100 employs an adaptive approach. It uses a set of bandpass filters in the digital domain to sample small chunks of bandwidth for short periods of time. It uses the frequency and these small chunks of bandwidth to determine whether it has captured the signal long enough to determine its configuration, or whether it needs to sample longer. The wireless network system 100 never uses more than two symbol periods in any configuration to make this determination. This concept is illustrated in FIG. 7.

[0030] Finally, to be compatible with existing hardware, the gateway must receive the entire packet after it has been configured. This goal seems difficult, if not impossible, because SDR uses at least a portion of the preamble to identify the gateway. One way to solve this problem is to buffer time samples in the rate-determining gateway 105 of the wireless network system 100 and then replay them in the base station 104 radio 106. However, this increases the circuit complexity and cost of the wireless network system 100. Instead, the preamble structure in the LoRa protocol is utilized to solve this problem. Critical to the operation of this system is the operational principle that the packet preamble length can be dynamically configured. A natural operational principle is that the dynamically configured preamble can be longer than the preamble length required by the base station radio to detect the packet. The remaining symbols can be used by the wireless network system 100 to determine configuration parameters and set these parameters in the base station radio. For example, a base station can be configured to expect an eight-symbol preamble, while a client can be configured to use ten symbols. These two extra symbols can be allocated for the purpose of allowing the base station of the wireless network system 100 to predict the coding parameters of the incoming packet transmission signal and reconfigure the LoRa base station to properly receive the signal based on the coding parameters. The gateway can then use the remaining signal to decode the packet. Note that because the number of up chirps is variable, the gateway can still recognize the complete preamble with a sequence of up chirps followed by two down chirps and successfully decode the packet.

[0031] Section 3: LoRa LoRa is a physical layer implementation for LPWAN based on chirp spread spectrum (CSS) technology. In LoRa modulation, chirp signals are generated for coded data symbols. The frequency of the chirp varies linearly with time as shown in Figure 8. Two parameters define the effective data rate: bandwidth and spreading factor. Bandwidth controls the total span of the chirp in the frequency domain. Spreading factor defines how long each chirp is in the time domain. Specifically, for a chirp with spreading factor SF, the time it takes to transmit is 2 SF is directly proportional to

[0032] Therefore, the time it takes to transmit a chirp, T s is T s =2 SF The bandwidth is given by / BW, where BW is the chirp and SF is the spreading factor. Thus, a higher bandwidth reduces the period of each chirp, and a higher spreading factor exponentially increases the period of each chirp.

[0033] To communicate a bit of information, the transmitter varies the initial frequency f of the chirp. Specifically, to transmit a symbol value S, the transmitter sets the starting frequency as follows:

[0034]

number

[0035]

number

[0036]

number

[0037] In conclusion, the terminology is repeated throughout the rest of the document: A symbol is the unit of data conveyed by each chirp. The period of a symbol is the same as the period of a chirp. Each symbol or chirp consists of multiple samples, depending on the sampling rate and sample period. For example, at 10 samples per second 6 For a sampling rate of 1000 samples, a symbol period of 2 milliseconds corresponds to 2000 samples.

[0038] Section 4: Wireless Network Systems The wireless network system 100 disclosed herein is a novel gateway design for LoRa that supports dynamic link configuration. The wireless network system 100 allows clients to optimize their data rates without having to notify the wireless base station 104 of these updated configurations. This, in turn, allows a single wireless base station 104 to support hundreds of wireless devices 101 on a large scale without compromising performance. For example, a LoRa network deployment may include client devices dispersed over a several-mile radius from a base station device 102. Over this coverage area, achievable throughput varies with distance and varying channel conditions. The wireless network system 100 enables a LoRa network to support a wide range of configurations that would otherwise require compromising performance to support all devices within a large coverage area.

[0039] To better understand LoRa performance, range tests were conducted to determine the maximum achievable data rate with respect to distance from the base station device 102. FIG. 10 shows a coverage map of the best configuration settings that can be supported while maintaining a reliable communication link between the LoRa base station and the client. In an industrial campus setting, the base station was placed in a fixed location, and client locations were varied throughout the campus. The client wireless devices continuously transmitted LoRa packets at 20 dB transmit power, varying the coding parameters at each location to test the system's limits. FIG. 10 shows the maximum data rate supported across all locations, along with the corresponding BW and SF. Importantly, there is a great deal of variation across the supported coding parameters, justifying the desire to support a more dynamic network.

[0040] The wireless network system 100 accomplishes this by employing a neural network approach to predict the bandwidth and spreading factor used by any given client for data transmission. The base station device's radio is then reconfigured accordingly to properly receive and decode incoming packets. A diagram of one architecture of a base station device 102 of the wireless network system 100 is shown in FIG. 5. As shown, the base station device 102 includes a transmission rate determination gateway 105. The transmission rate determination gateway 105 is an SDR, as described above, that has three components: a packet detection module 110 (packet detector) for detecting incoming LoRa packet transmissions; a classifier 112 (which may be a neural network processing unit, as described below) for classifying coding configurations; and finally, a radio configuration module 114 that communicates with the LoRaWAN wireless base station 104 to update the coding parameters. It is recognized that although the LoRaWAN wireless base station 104 includes "base station" in its name and the transmission rate determination gateway 105 includes "gateway" in its name, both are included in a single device that functions as a base station device 102 and also functions as a gateway to the WAN when connected to the WAN.

[0041] 1 shows at 100 a general diagram of the thus-described wireless network system in which a base station device 102 as described in FIG. 5 may be deployed. As shown, the wireless network system 100 includes a base station device 102 configured to communicate with a plurality of wireless devices 101 (e.g., LoRaWAN-configured devices) over a wireless network 108 (e.g., a LoRa network) using signals 107. The base station device 102 is configured to act as a gateway device to a wide area network (WAN) such as the Internet, over which the base station may communicate with removable devices such as, for example, a remote server and a remote client.

[0042] The base station device 102 includes a processing circuit 103 configured to detect a transmission rate from a portion of a preamble 107A1 of an incoming packet transmission signal 107A and adapt its radio 106 to receive the remainder 107A2 of the incoming packet transmission signal 107A at that transmission rate. The base station device 102 is configured to implement a low-power wide area network, and the incoming packet transmission signal 107A is transmitted from the wireless device 101 to the base station device 102 according to the LoRaWAN communication protocol. Thus, in this example, the incoming packet transmission signal 107A is transmitted from multiple wireless devices 101 using the LoRaWAN network protocol, although other network protocols may be used. For example, other low-power, long-range protocols may be used, or higher-speed network protocols such as 6G or other suitable network protocols may be used. In this example, three wireless devices 101 are shown communicating with the base station device 102, but it will be appreciated that up to several thousand wireless devices 101 may be communicating with the base station device 102.

[0043] Continuing with FIG. 1 , the transmission rate determination gateway 105 (which, as noted above, is an SDR) of the base station device 102 further includes a packet detection module 110 that implements an adaptive sampling algorithm to collect samples of the preamble 107A1 of an incoming packet transmission signal 107A being received by the receiver 115 of the base station device 102 from one of the plurality of wireless devices 101. The transmission rate determination gateway 105 of the base station device 102 further includes a classifier 112, which may be in the form of a CNN, configured to receive the samples and output a classification 117 indicative of one or more coding parameters of the incoming packet transmission signal. In this example, the coding coefficients are bandwidth and spreading factor, although other coding coefficients may be used in other examples. The transmission rate determination gateway 105 further includes a radio configuration module 114 that transmits configuration commands to configure the radio 106 of the wireless base station 104 to receive the remainder 107A2 of the incoming packet transmission signal 107A according to the one or more coding parameters, such as bandwidth and spreading factor, indicated by the classification 117. The process described in this paragraph is also illustrated in Figure 3, which shows that preamble 107A1 is processed by an adaptive sampling algorithm to generate samples corresponding to the first symbol of the preamble, which is then processed by classifier 112 to generate classifications 117 indicative of the coding parameters, which are then used to configure radio 106 to properly receive the remainder 107A2 of the incoming packet transmission signal 107A.

[0044] Three technical challenges exist for implementing the wireless network system 100. First, the wireless network system 100 faces the challenge of determining the configuration parameters of received packets in near real time. Second, the wireless network system 100 faces the challenge of being backward compatible with existing LoRa solutions. Third, the wireless base station device 102 of the wireless network system 100 faces the challenge of achieving high prediction accuracy across a variety of possible encoding parameters that may be selected by the wireless device 101. The following sections detail how the wireless network system 100 addresses each challenge and describe neural network architectures that may be used to implement the classifier 112.

[0045] 4.1 Real-time prediction To successfully decode the incoming packet transmission signal 107A, the base station device 102 needs to configure its radio 106 with parameters that match the incoming packet transmission signal 107A. This reconfiguration needs to be accomplished quickly enough so that the radio 106 still has time to detect the incoming packet transmission signal 107A. To detect the incoming packet transmission signal 107A, the radio 106 needs the preamble 107A1 of the incoming packet transmission signal 107A.

[0046] As mentioned in Section 2, the wireless network system 100 uses additional symbols added to the LoRa packet preamble 107A to determine configuration parameters and configure these parameters in the radio 106 of the base station device 102. To verify this technique, two Semtech SX 1276 LoRa chips were configured as the base station device 102 and the wireless device 101, respectively. The API of a common LoRa chipset (Semtech SX 1262 / 1276) was used to configure the LoRa packet preamble from 6 to 65,535 symbols. A minimum of six symbols is required for packet detection. The preamble 107A1 of the base station device 102 was configured to eight symbols, while the preamble length of the wireless device 101 was varied. The wireless device 101 transmitted packets over the air using different preamble lengths, and packet reception was then confirmed at the base station device 102. The results demonstrate that an additional five symbols can be added to the preamble 107A1 of the wireless device 101 while maintaining reliable reception at the base station device 102. The wireless network system 100 requires up to two symbols depending on the encoding parameters used for the input data. This variation is due to the fact that the input shape of the data is consistent for any given neural network. Because the duration of a symbol is a function of the BW and SF, the number of symbols used for any given input will also vary depending on the number of data samples passed to the network for classification.

[0047] 4.2 Estimation of SF and BW As noted above, the coding parameters are not pre-negotiated between the wireless device 101 and the base station device 102 before the base station device 102 receives the incoming packet transmission signal 107A. It is recognized that in the wireless network system 100, the wireless device 101 is configured to set coding parameters, such as bandwidth and spreading factor, to values ​​selected by the wireless device 101 from among multiple pre-set values ​​of the coding parameters. These pre-set values ​​typically include all possible values ​​defined as usable by a network protocol, such as LoRaWAN, and are not typically a subset of such possible coding parameters set during a configuration step by a network operator. Once the wireless device 101 autonomously selects the coding parameters, the wireless device 101 is configured to begin transmitting the incoming packet transmission signal 107A in accordance with the coding parameters without engaging in any prior communication with the base station device 102 to pre-negotiate the coding parameters.

[0048] The wireless network system 100 aims to predict the spreading factor and bandwidth of LoRa packet transmissions using neural network techniques. Before diving into the network architecture, we first explain why BW and SF can be estimated. The difference between specific combinations of BW and SF can be easily distinguished by simply comparing the number of samples per symbol. However, there are cases where the total number of samples matches (e.g., BW=125kHz, SF=8 and BW=500kHz, SF=10).

[0049] One approach to distinguishing between coding configurations is to first compare the frequency increase over time for any given chirp. This provides insight into the spreading factor. The start and stop frequencies of the chirp can then be used to determine the bandwidth. Referring again to Figures 8A and 8B, it can be seen that the rate of change over frequency varies with the spreading factor, and the difference between the start and stop frequencies yields the bandwidth used for the chirp. This technique is sufficient if the entire symbol period is used to predict the parameters, but since a single symbol can have a duration of as much as 525 milliseconds, doing so significantly increases the delay. Thus, the number of samples used to determine the coding parameters is minimized.

[0050] The above method can still be used to estimate the spreading factor and bandwidth using a subset of samples of the LoRa preamble symbols, but the tradeoff in this case is accuracy. Distinguishing between different coding parameters can be more difficult when considering the RSSI and SNR variations that a signal may experience when transmitting over the air. The wireless network system 100 takes into account the described characteristics of the LoRa chirp to train a convolutional neural network (CNN) to classify many different combinations of spreading factor and bandwidth. Specifically, three features extracted from the symbols of the LoRa preamble 107A1 are used to perform the classification:

[0051] 2, it is recognized that wireless network system 100 may include an analog-to-digital converter 111 configured to sample incoming transmission signals at various rates under the control of an adaptive sampling algorithm 113 implemented by packet detection module 110. Typically, samples from no more than two symbols are used by an artificial intelligence model of classifier 112, described in a following section, to output a classification 117. Thus, specifically, the samples include samples from two symbols (e.g., symbol (0) and symbol (1)) within preamble 107A1 of incoming packet transmission signal 107A, and the artificial intelligence model of classifier 112 determines classification 117 using multiple extracted features of the samples, the multiple features including a real component of the sample, an imaginary component of the sample, and a fast Fourier transform of the sample.

[0052] The first two features are the real and imaginary components of the signal, and the last is the Fast Fourier Transform (FFT). Using data from both the time and frequency domains of the signal is important in achieving high prediction accuracy. For example, if only the FFT of each signal is used, it is nearly impossible to distinguish between very low BW settings. As shown in Figure 9, when evaluating the FFT for different bandwidth and spreading factor settings, the lower kHz ranges begin to look quite similar. Complementing this with features from the time domain helps capture variations in the oscillation frequency of the preamble symbols, and the FFT provides insight into bandwidth variations.

[0053] 4.3 Adaptive Sampling The wireless network system 100 uses an adaptive sampling scheme to optimize sensitivity, delay, and classification accuracy. Referring again to Figure 7, the adaptive sampling method is illustrated. A digital bandpass filter is used to create a subset of the short-term bandwidth.

[0054] The adaptive sampling algorithm 113 shown in FIG. 2 is configured to filter the incoming packet transmission signal 107A using one or more bandpass filters, thereby generating a plurality of filtered incoming packet transmission signal components, and determine that the captured signal is sufficient to determine one or more coding parameters for one of the filtered incoming packet transmission signal components.

[0055] These subsets are used to determine whether the captured signal is long enough to provide accurate insight into the radio configuration or whether sampling should continue. In particular, the wireless network system 100 uses a total of 12,808 samples (65 ms) for the first six classes, which represent the two low bandwidths, and 800 samples (4 ms) for the last nine classes, which represent the high bandwidth radio configurations. Intuitively, since the symbol period increases as the BW decreases, it makes sense to use a larger set of samples for the low bandwidth configuration.

[0056] 4.4 Classifier Architecture The classifier 112 shown in FIG. 1 may be implemented as an artificial intelligence model including at least one convolutional neural network and may use a hierarchical neural network architecture including multiple stages, e.g., two stages. Thus, as shown in FIGS. 11-12B, the artificial intelligence model may be a multi-stage model, thus including a first stage and a second stage. The first stage may include a bandwidth classifier including a first convolutional neural network that classifies incoming packet transmission signals into one of multiple bandwidth range classifications, e.g., high and low. An intermediate range between the high and low bands may also be defined. In the second stage, for signals having a bandwidth below a predetermined threshold, the signals are classified into one of multiple low-bandwidth encoding classifications by a low-bandwidth encoding classifier including a second convolutional neural network, and for signals above the predetermined threshold, the signals are classified into one of multiple high-bandwidth encoding classifications by a high-bandwidth encoding classifier including a third convolutional neural network.

[0057] Continuing with the illustrated embodiment of Figure 11, a binary classifier is first used to distinguish between low and high bandwidth. Depending on the prediction, this is followed by a six or nine class classifier to predict the BW and SF radio configurations. Figure 12 shows the neural network architecture used by the wireless network system at each stage, with the main differences being the number of classes, features, and input samples per classifier.

[0058] The low-bandwidth classifier relies on the three features mentioned above. The binary and nine-class classifiers use 30 features to predict the radio configuration. The features include real and imaginary components and FFTs, while the samples are divided into ten 20 kHz chunks. The variation in the number of samples and features per classifier is selected based on the type of signal that needs to be classified, as explained above in Section 2. For example, 16 times more samples are used for the low-bandwidth classifier because symbol durations can be tens of milliseconds, requiring more samples to have meaningful features. On the other hand, the binary classifier only uses 800 samples even at low bandwidth, which is sufficient since there is no need to distinguish between individual bandwidths.

[0059] Each classifier neural network begins with four convolutional layers, each with a filter size of 128. The layers convolve the input and are activated by the ReLu (Rectified Linear Unit) function. The ReLu activation function outputs a maximum of 0 and the input data, providing output in the form of a feature map. Next is a max-pooling layer, which is used to reduce the size of the generated feature map and retain the most meaningful information. In this network, a max-pooling size of 2 is used. This is followed by six additional convolutional layers, each with a filter size ranging from 128 to 32. These layers also use the ReLu activation function. A global average pooling layer is added after this, calculating the average output of each feature map from the previous convolutional layer. A final densely connected layer is applied with a size equal to the total number of possible classifications. The densely connected layer uses a sigmoid activation function, which provides output probabilities across all classes between values ​​of 0 and 1. The maximum probability of the final output layer is used to obtain the predicted class.

[0060] To evaluate how well a neural network models a dataset, a categorical cross-entropy loss function is used.

[0061]

number

[0062] Three batch normalization and dropout layers are also added within the network. The batch normalization layer normalizes the output of the previous layer by subtracting the batch mean and dividing by the batch standard deviation. A batch is a portion of data passed to the model for training. Batch normalization helps improve network stability and reduces the number of epochs required to train the network. Finally, for normalization, a dropout of 0.5 is used before the final densely connected layer to reduce overfitting.

[0063] Section 5: Implementation Details regarding the implementation of the wireless network system and the setup for experimental evaluation are presented below.

[0064] 5.1 Hardware A hardware prototype of the gateway of wireless network system 100 is designed using a universal software radio (USRP) platform. The gateway of wireless network system 100 operates at 915 MHz, the frequency used by most LoRa deployments in the United States. The USRP is co-located with the LoRa receiver, which must be configured in its correct configuration to successfully receive packets.

[0065] The client is designed using the 1276 Semtech chipset. This chipset allows for spreading factors between 7 and 12 and bandwidths between 7.8 kHz and 500 kHz. Bandwidths of 10.4 kHz, 15.6 kHz, 125 kHz, 250 kHz, and 500 kHz were selected for the experiments to cover the extreme ends of the spectrum. By selecting two of the lowest possible bandwidths, it was recognized that the smallest difference between the bandwidths would be used. Finally, a spreading factor of 10-12 was used for the experiments.

[0066] The client chip is embedded in a PCB that sets the spreading factor, bandwidth, and allows data bits to be transmitted. The chip is controlled using an ARM STM32L151 microcontroller. Custom firmware is written for this microcontroller. The wireless network system 100 can operate without any modifications with any such implementation on the client side.

[0067] 5.2 Software The gateways of the wireless network system 100 are controlled using GNU Radio software. This software runs on a computer with 32 GB of RAM and collects samples at a center frequency of 915 MHz and a sampling rate of 200 ksps. This is the minimum sampling rate achievable by the USRP, resulting in a 200 kHz bandwidth at the receiver 115. Each packet recording is passed through a bandpass filter to further reduce the receiver bandwidth to 20 kHz. Additional filtering is performed to increase the sensitivity of the receiver 115. The samples are then separated into individual symbols using a packet detection algorithm that uses a combination of a sliding window power threshold and autocorrelation.

[0068] The CNN was implemented using the Tensorflow 2.0 framework in Python. It runs on an NVIDIA GeForce GTX 1050 GPU with 2GB of memory and a Microsoft Surface 2 with 16GB of RAM. The CNN was trained using the Adam optimizer with default parameters except for the learning rate, which was set to 0.0001. 20% of the training set was reserved as a validation set. The model was trained for 20 epochs in all experiments, and the best model was selected based on its performance on the validation set. Unless otherwise stated, each experiment was performed with three different training-test splits. The number of training points for each experiment is specified in the following section.

[0069] Section 6: Results An experimental evaluation of the wireless network system 100 is given below.

[0070] 6.1 Experimental setup To evaluate the wireless network system 100, a dataset was first generated to represent 15 possible classifications for spreading factors ranging from 10 to 12 and bandwidths of 10.4, 15.6, 125, 250, and 500 kHz. Because the LoRa packet preamble is a series of up-chirps, a dataset was created consisting of individual chirps in the form of complex baseband signals extracted from each packet preamble. The radio described in Section 5.1 was used to transmit and receive LoRa packets using USRP. In this setup, data was collected in a control setting, an indoor setting, and an outdoor setting.

[0071] Indoor Data Collection: The indoor experiments were conducted in an office space. The experiments spanned six different rooms covering a total area of ​​1000 square feet. The transmitting device (e.g., wireless device 101) and receiving device (e.g., base station device 102) were randomly placed in different rooms. In each configuration, data for each class was collected. On average, 800 symbols of data were collected per class per location.

[0072] Outdoor Data Collection: Data was collected using a campus-wide deployment to emulate an outdoor deployment. The receiving device was placed at a fixed location on the ground. The transmitting device was moved manually or on a vehicle to different locations within a campus area spanning 0.02 square miles. For each location, a random spreading factor and random bandwidth were selected to transmit the data. The GPS coordinates of the location and the settings used were manually recorded. Data was collected for a total of 16 locations on the campus.

[0073] Benchtop Data Collection: A benchtop experimental setup was used to create a control data set with various RSSIs (Receiver Signal Strength Indicators) to replicate the long-distance field experiments. In this setup, the transmitting and receiving devices were directly wired. A variable attenuator was used to attenuate the transmitted signal, using an attenuation range of 40 to 140 dB for each symbol classification.

[0074] Baseline: A baseline based on cross-correlation calculations was used. An exemplary sample set containing one sample signal per class (bandwidth and spreading factor pair) was used. For a given signal input S, f S,Ei (n) is the instance E in class i i is the cross-correlation of S with Then the similarity score for class i was calculated as follows:

[0075]

number

[0076] 6.2 Accuracy evaluation First, the accuracy of the CNN in the wireless network system 100 in identifying the correct configuration of a packet was evaluated. As described above, in the CNN in the wireless network system, the raw signal is captured for 4 milliseconds and used as input for a binary classifier. If the received packet is in the low-bandwidth category, the signal capture is increased to 65 milliseconds; otherwise, it remains the same for high-bandwidth data rates. This corresponds to two chirp (or symbol) periods for the highest data rate in the experiment (500 kHz bandwidth, 10-spreading factor) and approximately one-sixth of a chirp period for the lowest data rate. The performance of the neural network was evaluated by analyzing the accuracy of all three scenarios described above. Due to limitations, the analysis used a combination of indoor and benchtop data to train the network. 30% of the collected data was used for training, and all other data was used for testing.

[0077] Next, the accuracy of the wireless network system 100 will be described with reference to FIG. 13. As shown, the CNN of the wireless network system 100 achieves very high overall accuracy of 99.8%, 95%, and 98.2% for indoor, outdoor, and benchtop evaluations, respectively. This high accuracy demonstrates the feasibility of the main concept of the wireless network system 100, namely, that the correct configuration of packets can be identified with high accuracy at the gateway. In comparison, the performance of the baseline is significantly worse. In the three settings, the baseline accuracies are 67.5%, 67%, and 78%, respectively. One reason for the poor performance of the baseline is the challenge of identifying small differences in frequency bandwidths such as 10.4 kHz and 15.6 kHz. Unlike higher bandwidths such as 125 kHz and 250 kHz, these bandwidths are relatively close, and the presence of noise and multipath makes them difficult to distinguish.

[0078] Environmental fluctuations Figure 13 also shows the environmental variations. The system performs better outdoors than indoors. This is mainly because the outdoor environment comprises more free space and less multipath fading compared to the indoor environment. On the other hand, the indoor environment has much more multipath reflections, making it more difficult.

[0079] Variation with bandwidth 14A shows the performance of the wireless network system 100 across different bandwidths. In this experiment, the more meaningful recall is reported. Recall is the number of points correctly classified into bandwidth B divided by the number of points actually transmitted in bandwidth B. As shown, recall remains around 99% for all bandwidths, ranging from a low of 98.4% (for 15.6 kHz) to a high of nearly 100% for 10.4 kHz and 125 kHz.

[0080] Variation due to diffusion coefficient FIG. 14B shows the performance variation of the wireless network system 100 at different spreading factors. As shown, the recall remains around 99% for all three spreading factors. The recall is slightly lower for the highest spreading factor. This is mainly because the highest spreading factor corresponds to the maximum duration of each chirp. This means that if sampling is performed at a fixed period, as in the case of the input, the smallest percentage of chirps is obtained for the highest spreading factor. This makes the classification problem more difficult as the spreading factor increases. Nevertheless, the wireless network system achieves an accuracy of over 95% even for the highest spreading factor used by LoRa by using less than half of a single chirp period. This demonstrates the strong performance of the CNN design of the wireless network system.

[0081] Variation due to position 14C shows the performance variation of the wireless network system 100 in different physical spaces. L0 to L4 indicate four different locations. At each of these locations, the accuracy of the wireless network system 100 is consistent at approximately 99 to 100%.

[0082] Time Variation Figure 14D shows the performance variation over time of the wireless network system 100 for all 15 classes. In this experiment, 30 minutes of data were collected wirelessly for five consecutive days. As shown, the accuracy remains around 100% for all days. Compared to the baseline approach, the accuracy is significantly reduced to around 88%.

[0083] A key finding from the accuracy analysis is that the wireless network system 100 can correctly identify radio configurations in a diverse set of scenarios with high accuracy. The wireless network system achieved an overall accuracy of 97.7%, which corresponds to a packet loss of less than 1 / 20. This loss becomes insignificant when considering the overall packet loss of LoRa. Packet loss for a 125 kHz bandwidth and a spreading factor of 12 can range from 12% to 74% from a distance of 0 to 15 km in an outdoor urban scenario. The additional loss introduced by the wireless network system 100 is believed to be a reasonable trade-off to enable automatic radio configuration.

[0084] 6.3 Generalization One question that arises in most machine learning frameworks is their ability to generalize to new environments not seen in the training set. This issue is addressed using two experimental evaluations on wireless network systems.

[0085] First, the model is trained while excluding two locations (different rooms in the indoor environment) from the training data. Specifically, data obtained from L5 and L6 are excluded from the training set. The data from these two locations are set separately for the test set. This allows testing of generalization to new environments. The results of this experiment are shown in Figure 15A. As shown, the location accuracy suffers a slight decrease from 98.9% to 94.5%.

[0086] Next, the model was tested for generalization over time. Test data was collected on dates not included in the training set (set one week apart). The model maintained the performance (97% accuracy) achieved on the previous date. This indicates that while there was some location-to-location variation in accuracy, no temporal variation was observed. The main takeaway from these results is that the wireless network system can achieve high accuracy even for input signals from scenarios that the CNN has not encountered. This indicates that the CNN of the wireless network system can be used for a diverse set of LoRa networks.

[0087] 6.4 Sensitivity LoRa can operate with a sensitivity range of -149 to -118 dBm, depending on the SF and BW settings used. For the wireless network system 100 to be useful for LoRa network deployment, it must be able to achieve high accuracy over the same sensitivity range. To evaluate the accuracy of the CNN of the wireless network system 100 for signals with low power, attenuated data sets from 40 to 140 dB were generated and the model accuracy was analyzed. Figure 15B shows the model accuracy as a function of attenuation for the wireless network system and the baseline method. The wireless network system 100 has an average accuracy of 96.7% and a maximum of 99%, regardless of attenuation. This is consistent with the accuracy achieved in the overall benchtop experiments reported in Figure 13. On the other hand, the accuracy of the baseline method fluctuates and decreases for signals exposed to large amounts of attenuation. The overall results indicate that the wireless network system 100 is robust to signal strength variations and should be able to maintain prediction accuracy for signal conditions that LoRa may encounter.

[0088] 6.5 Delay Minimizing the latency of the wireless network system 100 is important for maintaining real-time predictions. As described above, it was determined that five additional symbols could be added to the preamble 107A1 of the LoRa packet transmission that can be allocated to the wireless network system 100 to detect, classify, and update the radio configuration at the base station device 102. This corresponds to a duration ranging from 0.01 seconds to 1.92 seconds. The wireless network system 100 uses a maximum of two symbols per class (less than one symbol for most classes), allowing the remaining time for classification and parameter configuration. The latency of the CNN of the wireless network system 100 was evaluated and compared with a baseline method. Figure 13C shows a latency comparison between the two methods. It can be seen that to perform classification, the wireless network system 100 requires approximately 60 milliseconds per sample using a CPU, while using a notebook NVIDIA GTX 1050 GPU results in a 20x latency improvement of approximately 3 milliseconds per sample. The baseline method has a computation time of 140 ms per sample, making it impossible to classify in real time for most LoRa coding parameter configurations.

[0089] Section 7: System Overview Described in this disclosure is a novel gateway design that enables wireless devices 101 to transmit at a data rate of their choice using LoRa and other protocols. This allows base station devices 102 to support, for example, a large number of mobile wireless devices 101 over long distances without compromising overall network performance. The wireless network system 100 uses CNNs to predict the bandwidth and spreading factor of packets transmitted by the wireless devices 101, enabling the base station devices 102 to decode packets across various signal coding parameter settings, thereby quickly configuring the radio 106 of the base station device 102 to correctly receive the remainder 107A2 of an incoming packet transmission signal 107A based solely on information from the first two symbols from the preamble 107A1.

[0090] The test implementation of the wireless network system 100 includes the following component functions:

[0091] LoRa wireless configuration classifier Test results show that the neural network implemented can classify 15 different LoRa wireless configurations with 99.8% and 95% accuracy in indoor and outdoor scenarios.

[0092] Real-time classification Testing has shown that by leveraging the dynamic preamble setting of LoRa packets, radio configuration can be automated and performed in real time. The wireless network system 100 relies on up to two preamble symbols to perform classification with high accuracy across a diverse set of scenarios.

[0093] Adaptive Sampling Adaptive sampling is implemented to optimize the tradeoff between network sensitivity, accuracy, and latency. The wireless network system 100 adapts bandwidth and capture period to cater to the vast set of radio configurations supported by LoRa.

[0094] Although a particular application of the disclosed wireless network system 100 is described herein, it is recognized that the wireless network system may be used in other applications, examples of which are described below.

[0095] Rate Adaptation The wireless network system 100 can be used to improve LoRa's rate adaptation techniques. Because clients can configure their own coding parameters and the wireless network system 100 can automatically configure the base station device 102 accordingly, much of the typical overhead can be avoided. For example, control messages between the base station and clients can be minimized. Developing new protocols built into the wireless network system 100 for rate adaptation has the potential to further improve the performance and efficiency of LPWANs.

[0096] Field Programmable Gate Array Implementation Although not shown, the wireless network system 100 can also be implemented in a field-programmable gate array (FPGA). FPGAs offer faster performance compared to other hardware computing platforms and also provide the flexibility to support different algorithms, logic, and memory resources. Such an implementation can help improve the latency of the wireless network system 100 by minimizing the time required to detect, classify, and update radio parameters.

[0097] Alternative Hardware Although the described system is developed as a gateway augmented with software-defined radio, several off-the-shelf gateways, such as the SX1257, support access to the raw IQ samples of the signal and are compatible with the design.

[0098] Network Pruning In relation to improving delay, network pruning used by the wireless network system 100 is a promising approach. The idea behind network pruning is that with many parameters in the network, there are bound to be some that are redundant and contribute little. This minimizes the size of the network, which in turn optimizes the time required to perform classification.

[0099] As 5G standardization reaches its final stages, there is growing interest in defining 6G networks, aiming to provide an order-of-magnitude improvement in bandwidth and latency compared to 5G. A promising approach being explored is machine learning, which automatically reconfigures devices to communicate with each other, including those using other standards. This can significantly reduce control overhead and result in increased network capacity. The architecture of wireless network system 100 is a step toward this vision of full interoperability while still maintaining backward compatibility with legacy devices. Therefore, the systems and methods described herein are believed to be applicable to future protocols, including future high-speed wireless communication protocols such as emerging 6G protocols.

[0100] A wireless network method will now be described with reference to Figure 16A. A wireless network method 1600 is provided. As shown, at 1602, in one embodiment, the method includes detecting a transmission rate from a portion of a preamble of an incoming packet transmission signal, and at 1614, the method includes adapting a radio to receive the remainder of the incoming packet transmission signal at that transmission rate. Further details of the method are provided below.

[0101] The method further includes, at 1604, implementing, via the processing circuitry, an adaptive sampling algorithm to collect samples of the preamble of the incoming packet transmission signal. The processing circuitry may be included in a base station having a radio configured to receive and transmit wireless signals. In this embodiment, the wireless signals are received and transmitted according to the LoRa network protocol, although other network protocols may be used in other embodiments. For example, other suitable low-power or long-range network protocols in which the length of the data symbols in the transmission signal varies significantly may benefit from application of this method. The incoming packet transmission signal is received from a wireless device. While the method of this embodiment describes an incoming packet transmission signal received from one wireless device, it will be appreciated that the method is also suitable for receiving incoming packet transmission signals from multiple wireless devices. For example, tens, hundreds, or even thousands of wireless devices may be used.

[0102] At 1606, the method further includes receiving the samples at a classifier and outputting a classification indicative of one or more coding parameters of the incoming packet transmission signal. The coding parameters are not pre-negotiated between the wireless device and the base station prior to receiving the incoming packet transmission signal. An advantage of not pre-negotiating the coding parameters is that the client device transmitting the incoming packet transmission signal can be used as is. In other words, the method described herein does not require modification of the client device. In this method, the one or more coding parameters include bandwidth and / or spreading factor, although other suitable coding parameters may be used.

[0103] At 1608, in one exemplary configuration of the method, the classifier is an artificial intelligence model that includes at least one convolutional neural network, details of which are shown in Figure 16B and described below.

[0104] In 1610, the samples include samples taken from two symbols in the preamble of the packet signal, and the classifier's artificial intelligence model determines the classification using multiple features of the samples, including the real component of the sample, the imaginary component of the sample, and a fast Fourier transform (FFT) of the sample. Using data from both the time and frequency domains of the signal is important to achieving high prediction accuracy. For example, if only the FFT of each signal is used, it is difficult to distinguish between very low BW settings. Supplementing the FFT with features from the time domain helps capture variations in the oscillation frequency of the preamble symbols, and the FFT helps provide insight into bandwidth variations. By using features from both the time and frequency domains of the signal, samples taken from no more than two symbols are used by the artificial intelligence model to output the classification.

[0105] At 1616, the method includes transmitting a configuration command to configure the radio to receive the remainder of the incoming packet transmission signal in accordance with the one or more coding parameters indicated by the classification, such that the remainder of the incoming packet transmission signal can be received by the radio.

[0106] 16B, further details of 1608 are provided. At 1618, the artificial intelligence model is a multi-stage model and includes a first stage, where a bandwidth classifier includes a first convolutional neural network that classifies the incoming packet transmission signal into one of a plurality of bandwidth range classifications. In this example, the bandwidth classifier including the first convolutional neural network uses the real component, the imaginary component, and an FFT of the incoming packet transmission signal, dividing each into ten 20 kHz chunks. However, in other examples, 2, 4, 6, 8, or any other suitable number of chunks may be used. In this example, the first stage classifies the incoming packet transmission signal into one of two bandwidth range classifications, although 3, 4, or any other suitable number may be used.

[0107] At 1620, in a second stage, for signals having a bandwidth below a predetermined threshold, the signal is classified into one of a plurality of low-bandwidth coding classifications by a low-bandwidth coding classifier including a second convolutional neural network.

[0108] At 1622, for signals above a predetermined threshold, the signal is classified into one of a plurality of high-bandwidth coding classifications by a high-bandwidth coding classifier including a third convolutional neural network.

[0109] In some embodiments, the methods and processes described herein may be coupled to the computing system of one or more computing devices. In particular, such methods and processes may be implemented as a computer application program or service, an application programming interface (API), a library, and / or other computer program product.

[0110] 17 illustrates generally a non-limiting embodiment of a computing system 1700 capable of performing one or more of the methods and processes described above. The computing system 1700 is illustrated in simplified form. The computing system 1700 may embody the wireless device 101, the base station device 102, and / or the remote device shown in FIG. 1 above. The computing system 1700 may take the form of one or more personal computers, server computers, tablet computers, home entertainment computers, network computing devices, gaming devices, mobile computing devices, mobile communication devices (e.g., smartphones), IoT devices, remote sensor devices, and / or other computing devices.

[0111] Computing system 1700 includes a logic processor 1702, a volatile memory 1704, and a non-volatile storage device 1706. Computing system 1700 may optionally include a display subsystem 1708, an input subsystem 1710, a communication subsystem 1712, and / or other components not shown in FIG.

[0112] Logic processor 1702 includes one or more physical devices configured to execute instructions. For example, logic processor 1702 may be configured to execute instructions that are part of one or more applications, programs, routines, libraries, objects, components, data structures, or other logical entities. Such instructions may be implemented to perform a task, implement a data type, transform the state of one or more components, achieve a technical effect, or arrive at a desired result in some other way.

[0113] A logic processor may include one or more physical processors (hardware) configured to execute software instructions. Additionally or alternatively, a logic processor may include one or more hardware logic circuits or firmware devices configured to execute hardware-implemented logic or firmware instructions. The processors of logic processor 1702 may be single-core or multi-core, and the instructions executed thereon may be configured for sequential, parallel, and / or distributed processing. Individual components of the logic processor may optionally be distributed across two or more separate devices, which may be remotely located and / or configured for cooperative processing. Aspects of the logic processor may be virtualized and executed by remotely accessible, networked computing devices arranged in a cloud computing configuration. In such cases, it is understood that these virtualized aspects may be executed on different physical logic processors on various different machines.

[0114] Non-volatile storage device 1706 includes one or more physical devices configured to hold instructions executable by a logic processor to implement the methods and processes described herein. When such methods and processes are implemented, the state of non-volatile storage device 1706 may be transformed, for example, to hold different data.

[0115] The non-volatile storage device 1706 may include removable and / or internal physical devices. The non-volatile storage device 1706 may include optical memory (e.g., CD, DVD, HD-DVD, Blu-Ray Disc, etc.), semiconductor memory (e.g., ROM, EPROM, EEPROM, flash memory, etc.), and / or magnetic memory (e.g., hard disk drive, floppy disk drive, tape drive, MRAM, etc.) or other mass storage device technologies. The non-volatile storage device 1706 may include non-volatile, dynamic, static, read / write, read-only, sequential access, location-addressable, file-addressable, and / or content-addressable devices. It will be appreciated that the non-volatile storage device 1706 is configured to retain instructions even when power to the non-volatile storage device 1706 is disconnected.

[0116] Volatile memory 1704 may include physical devices including random access memory. Volatile memory 1704 is typically utilized by logic processor 1702 to temporarily store information during the processing of software instructions. It is recognized that volatile memory 1704 typically does not continue to store instructions when power to volatile memory 1704 is disconnected.

[0117] Aspects of logic processor 1702, volatile memory 1704, and non-volatile storage device 1706 may be integrated together into one or more hardware logic components, which may include, for example, a field-programmable gate array (FPGA), a program- and application-specific integrated circuit (PASIC / ASIC), a program- and application-specific standard product (PSSP / ASSP), a system-on-chip (SOC), or a complex programmable logic device (CPLD).

[0118] The terms “module,” “program,” and “engine” may be used to describe aspects of computing system 1700 that are typically implemented in software by a processor using portions of volatile memory to perform a particular function, which function includes transformations that specifically configure the processor to perform the function. Thus, a module, program, or engine may be instantiated using portions of volatile memory 1704 via logic processor 1702 executing instructions maintained by non-volatile storage device 1706. It will be understood that different modules, programs, and / or engines may be instantiated from the same application, service, code block, object, library, routine, API, function, etc. Similarly, the same module, program, and / or engine may be instantiated by different applications, services, code blocks, objects, routines, APIs, functions, etc. The terms “module,” “program,” and “engine” may include individual or groupings of executable files, data files, libraries, drivers, scripts, database records, etc.

[0119] If included, the display subsystem 1708 may be used to present a visual representation of the data maintained by the non-volatile storage device 1706. The visual representation may be in the form of a graphical user interface (GUI). As the methods and processes described herein modify the data maintained by the non-volatile storage device and thus transform the state of the non-volatile storage device, the state of the display subsystem 1708 may likewise be transformed to visually represent the changes in the underlying data. The display subsystem 1708 may include one or more display devices utilizing virtually any type of technology. Such display devices may be combined with the logic processor 1702, the volatile memory 1704, and / or the non-volatile storage device 1706 in a shared housing, or such display devices may be peripheral display devices.

[0120] If included, input subsystem 1710 may include or interface with one or more user input devices such as a keyboard, mouse, camera, microphone, touchpad, finger-operable pointer device, touchscreen, or game controller.

[0121] Communications subsystem 1712, if included, may be configured to communicatively couple the various computing devices described herein with each other or with other devices. Communications subsystem 1712 may include wired and / or wireless communication devices compatible with one or more different communications protocols, including low-power, long-range wireless protocols such as LoRaWAN described above. As non-limiting examples, communications subsystem may be configured for communication over a wireless telephone network, a wired or wireless local or wide area network, or a similar network. In some embodiments, communications subsystem may enable computing system 1700 to send and / or receive messages to and from other devices over a network such as the Internet.

[0122] The following paragraphs provide further description of the subject matter of this disclosure: According to one aspect, there is provided a wireless network system including a base station device including processing circuitry configured to detect a transmission rate from a portion of a preamble of an incoming packet transmission signal and adapt a radio configuration to receive the remainder of the incoming packet transmission signal at the transmission rate.

[0123] In this aspect, the base station device may further include a packet detection module implementing an adaptive sampling algorithm to collect samples of a preamble of an incoming packet transmission signal. The incoming packet transmission signal is received from a wireless device by a receiver of the base station device. The base station device may further include a classifier configured to receive the samples and output a classification indicative of one or more coding parameters of the incoming packet transmission signal. The base station device may further include a radio configuration module that transmits configuration commands to configure a radio of the base station device to receive the remainder of the incoming packet transmission signal according to the one or more coding parameters indicated by the classification.

[0124] In this aspect, the coding parameters do not have to be pre-negotiated between the wireless device and the base station device prior to receiving the incoming packet transmission signal.

[0125] In this aspect, the wireless device may be further configured to set the coding parameter to a value selected at the wireless device from among a plurality of pre-configured values ​​of the coding parameter and to commence transmission of an incoming packet transmission signal in accordance with the coding parameter without engaging in prior communication with the base station device to pre-negotiate the coding parameter.

[0126] In this aspect, the samples may include samples taken from two symbols in a preamble of the packet signal, and the artificial intelligence model of the classifier determines the classification using multiple features of the samples, the multiple features including a real component of the samples, an imaginary component of the samples, and a fast Fourier transform of the samples.

[0127] In this aspect, samples taken from no more than two symbols may be used to output a classification by the artificial intelligence model.

[0128] In this aspect, the one or more coding parameters may include a bandwidth and / or a spreading factor.

[0129] In this aspect, the adaptive sampling algorithm may be further configured to filter the incoming packet transmission signal using one or more bandpass filters, thereby generating a plurality of filtered incoming packet transmission signal components, and determine that the captured signal is sufficient to determine one or more coding parameters for one of the filtered incoming packet transmission signal components.

[0130] In this aspect, the classifier may include an artificial intelligence model including at least one convolutional neural network.

[0131] In this aspect, the artificial intelligence model may be a multi-stage model, including a first stage in which a bandwidth classifier includes a first convolutional neural network that classifies incoming packet transmission signals into one of a plurality of bandwidth range classifications, and a second stage in which, for signals having a bandwidth below a predetermined threshold, the signal is classified into one of a plurality of low-bandwidth encoding classifications by a low-bandwidth encoding classifier including a second convolutional neural network, and for signals above the predetermined threshold, the signal is classified into one of a plurality of high-bandwidth encoding classifications by a high-bandwidth encoding classifier including a third convolutional neural network.

[0132] In this aspect, the base station device may be configured to implement a low-power wide area network, and incoming packet transmission signals are transmitted from the wireless device to the base station device according to a LoRaWAN communication protocol.

[0133] According to another aspect, a wireless network method is provided that includes detecting a transmission rate from a portion of a preamble of an incoming packet transmission signal and adapting a radio to receive the remainder of the incoming packet transmission signal at that transmission rate.

[0134] In this aspect, the method may further include implementing, via the processing circuitry, an adaptive sampling algorithm to collect samples of a preamble of an incoming packet transmission signal received from the wireless device; receiving the samples in a classifier and outputting a classification indicative of one or more coding parameters of the incoming packet transmission signal; and transmitting a configuration command to configure the radio to receive the remainder of the incoming packet transmission signal according to the one or more coding parameters indicated by the classification.

[0135] In this aspect, the coding parameters do not have to be pre-negotiated between the wireless device and the base station device prior to receiving the incoming packet transmission signal.

[0136] In this aspect, the samples may include samples taken from two symbols in a preamble of the packet signal, and the artificial intelligence model of the classifier determines the classification using multiple features of the samples, the multiple features including a real component of the samples, an imaginary component of the samples, and a fast Fourier transform of the samples.

[0137] In this aspect, the samples may be taken from no more than two symbols used to output a classification by the artificial intelligence model.

[0138] In this aspect, the one or more coding parameters may include a bandwidth and / or a spreading factor.

[0139] In this aspect, the classifier may be an artificial intelligence model that includes at least one convolutional neural network.

[0140] In this aspect, the artificial intelligence model may be a multi-stage model, including a first stage in which a bandwidth classifier includes a first convolutional neural network that classifies incoming packet transmission signals into one of a plurality of bandwidth range classifications, and a second stage in which, for signals having a bandwidth below a predetermined threshold, the signal is classified into one of a plurality of low-bandwidth encoding classifications by a low-bandwidth encoding classifier including a second convolutional neural network, and for signals above the predetermined threshold, the signal is classified into one of a plurality of high-bandwidth encoding classifications by a high-bandwidth encoding classifier including a third convolutional neural network.

[0141] According to another aspect, a wireless network system is provided that includes processing circuitry configured to execute a packet detection module that implements an adaptive sampling algorithm to collect samples of a preamble of an incoming packet transmission signal, the incoming packet transmission signal being received by a receiver from a wireless device. The wireless network system may be further configured to execute a classifier that includes a neural network configured to receive the samples and output a classification indicative of one or more coding parameters of the incoming packet transmission signal. The wireless network system may be further configured to execute a radio configuration module that transmits configuration commands to configure an associated radio to receive the remainder of the incoming packet transmission signal according to the one or more coding parameters indicated by the classification.

[0142] It is understood that the structures and / or techniques described herein are exemplary in nature and are susceptible to numerous variations, and therefore, these specific embodiments or examples are not to be considered limiting. A particular routine or method described herein may represent one or more of any number of processing strategies. As such, the various operations illustrated and / or described may be performed in the order illustrated and / or described, in other orders, in parallel, or omitted. Similarly, the order of the processes described above may be changed.

[0143] The subject matter of the present disclosure includes all novel and non-obvious combinations and subcombinations of the various processes, systems, and configurations, as well as other features, functions, operations and / or properties disclosed herein, and all equivalents thereof.

[0144] The following concepts are further provided for the present disclosure.

[0145] (1) A wireless network system, 1. A wireless network system including a base station device including processing circuitry configured to detect a transmission rate from a portion of a preamble of an incoming packet transmission signal and to adapt a radio configuration to receive the remainder of the incoming packet transmission signal at the transmission rate.

[0146] (2) the base station device, a packet detection module that implements an adaptive sampling algorithm to collect samples of the preamble of the incoming packet transmission signal, the incoming packet transmission signal being received from a wireless device by a receiver of the base station device; and a classifier configured to receive the samples and output a classification indicative of one or more coding parameters of the incoming packet transmission signal; a radio configuration module that transmits configuration commands to configure the radio of the base station device to receive the remainder of the incoming packet transmission signal in accordance with the one or more encoding parameters indicated by the classification; The wireless network system according to (1), further comprising:

[0147] (3) The wireless network system of (2), wherein the coding parameters are not pre-negotiated between the wireless device and the base station device prior to receiving the incoming packet transmission signal.

[0148] (4) the wireless device, setting the encoding parameter to a value selected at the wireless device from among a plurality of preset values ​​of the encoding parameter; and commencing transmission of the incoming packet transmission signal in accordance with the coding parameters without engaging in prior communication with the base station device to pre-negotiate the coding parameters. The wireless network system according to (2), further configured as follows:

[0149] (5) The wireless network system described in (2), wherein the samples include samples obtained from two symbols in a preamble of the packet signal, and the artificial intelligence model of the classifier determines the classification using multiple features of the samples, the multiple features including a real component of the samples, an imaginary component of the samples, and a fast Fourier transform of the samples.

[0150] (6) The wireless network system of (5), wherein samples taken from no more than two symbols are used by the artificial intelligence model to output the classification.

[0151] (7) The wireless network system described in (2), wherein the one or more coding parameters include a bandwidth and / or a spreading factor.

[0152] (8) The adaptive sampling algorithm filtering the incoming packet transmission signal using one or more bandpass filters, thereby generating a plurality of filtered incoming packet transmission signal components; determining that the captured signal is sufficient to determine the one or more coding parameters for one of the filtered incoming packet transmission signal components; The wireless network system according to (2), configured as follows:

[0153] (9) The wireless network system described in (2), wherein the classifier is an artificial intelligence model including at least one convolutional neural network.

[0154] (10) The artificial intelligence model is a multi-stage model; a first stage, wherein a bandwidth classifier includes a first convolutional neural network that classifies the incoming packet transmission signal into one of a plurality of bandwidth range classifications; a second stage in which, for signals having a bandwidth below a predetermined threshold, the signals are classified into one of a plurality of low-bandwidth coding classes by a low-bandwidth coding classifier comprising a second convolutional neural network, and, for signals above the predetermined threshold, the signals are classified into one of a plurality of high-bandwidth coding classes by a high-bandwidth coding classifier comprising a third convolutional neural network; The wireless network system according to (5), comprising:

[0155] (11) The wireless network system described in (1), wherein the base station device is configured to implement a low-power wide area network, and the incoming packet transmission signal is transmitted from the wireless device to the base station device according to a LoRaWAN communication protocol.

[0156] (12) A wireless network method, comprising: detecting a transmission rate from a portion of a preamble of an incoming packet transmission signal; adapting the radio to receive the remainder of the incoming packet transmission signal at said transmission rate; A method comprising:

[0157] (13) via a processing circuit, implementing an adaptive sampling algorithm to collect samples of the preamble of the incoming packet transmission signal received from a wireless device; receiving the samples in a classifier and outputting a classification indicative of one or more coding parameters of the incoming packet transmission signal; transmitting a configuration command to configure the radio to receive the remainder of the incoming packet transmission signal in accordance with the one or more coding parameters indicated by the classification; The method according to (12), further comprising:

[0158] (14) The method of (13), wherein the coding parameters are not pre-negotiated between the wireless device and the base station device prior to receiving the incoming packet transmission signal.

[0159] (15) The method of (13), wherein the samples include samples taken from two symbols in a preamble of the packet signal, and the artificial intelligence model of the classifier determines the classification using multiple features of the samples, the multiple features including a real component of the samples, an imaginary component of the samples, and a fast Fourier transform of the samples.

Claims

1. 1. A wireless network system including a processing circuit, The processing circuitry a classifier configured to receive samples of preambles of incoming packets of an incoming packet transmission signal and to output a classification indicative of one or more coding parameters of the incoming packet transmission signal; a radio configuration module that transmits configuration commands to configure a radio to receive the remainder of the incoming packet transmission signal in accordance with the one or more encoding parameters indicated by the classification; configured to run the incoming packet transmission signal is received from a wireless device; The one or more coding parameters are not pre-negotiated between the wireless device and the processing circuitry prior to receiving the incoming packet transmission.

2. 2. The wireless network system of claim 1, wherein the processing circuitry is configured to detect a transmission rate from a portion of the preamble of the incoming packet transmission signal and adapt a radio configuration to receive the remainder of the incoming packet transmission signal at the transmission rate.

3. 10. The wireless network system of claim 1, wherein the processing circuitry is further configured to execute a packet detection module that implements an adaptive sampling algorithm for collecting the samples of the preamble of the incoming packet transmission signal.

4. The adaptive sampling algorithm comprises: filtering the incoming packet transmission signal using one or more bandpass filters, thereby generating a plurality of filtered incoming packet transmission signal components; determining that the signal is sufficient to determine the one or more coding parameters for one of the plurality of filtered incoming packet transmission signal components; 4. The wireless network system according to claim 3, configured as follows:

5. The wireless device setting the encoding parameter to a value selected at the wireless device from among a plurality of preset values ​​of the encoding parameter; Initiating transmission of the incoming packet transmission signal in accordance with the coding parameters without engaging in prior communication with the processing circuitry to pre-negotiate the coding parameters. The wireless network system according to claim 1 , configured as follows:

6. the processing circuitry is configured to implement a low power wide area network; 10. The wireless network system of claim 1, wherein the incoming packet transmission signal is received by the processing circuitry according to a Long Range Wide Area Network (LoRaWAN) communication protocol.

7. 2. The wireless network system of claim 1, wherein the artificial intelligence model of the classifier determines the classification using a plurality of features of the sample, the plurality of features including a real component of the sample, an imaginary component of the sample, and a fast Fourier transform of the sample.

8. 10. The wireless network system of claim 1, wherein the classifier is an artificial intelligence model including at least one convolutional neural network.

9. The artificial intelligence model is a multi-stage model, a first stage, wherein a bandwidth classifier includes a first convolutional neural network that classifies the incoming packet transmission signal into one of a plurality of bandwidth range classifications; a second stage in which, for signals having a bandwidth below a predetermined threshold, the signals are classified into one of a plurality of low-bandwidth coding classes by a low-bandwidth coding classifier comprising a second convolutional neural network, and, for signals above the predetermined threshold, the signals are classified into one of a plurality of high-bandwidth coding classes by a high-bandwidth coding classifier comprising a third convolutional neural network; 9. The wireless network system of claim 8, comprising:

10. The wireless network system of claim 1 , wherein the one or more coding parameters include a bandwidth and / or a spreading factor.

11. 1. A wireless network system including a base station device including processing circuitry, The processing circuitry of the base station device a packet detection module implementing an adaptive sampling algorithm for collecting samples of a preamble of an incoming packet transmission signal, the incoming packet transmission signal being received by a receiver of the base station device from a wireless device; a bandwidth classifier including a first convolutional neural network that classifies the incoming packet transmission signal into one of a plurality of bandwidth range classifications; a low-bandwidth coding classifier including a second convolutional neural network that classifies signals classified by the bandwidth classifier as having a bandwidth below a predetermined threshold into one of a plurality of low-bandwidth coding classifications; a high-bandwidth coding classifier including a third convolutional neural network that classifies signals classified by the bandwidth classifier as having a bandwidth above the predetermined threshold into one of a plurality of high-bandwidth coding classifications; a radio configuration module that transmits configuration commands to configure the radio of the base station device to receive the remainder of the incoming packet transmission signal in accordance with one or more encoding parameters indicated by the classification; A wireless network system comprising:

12. 1. A wireless network method, comprising: receiving, at a classifier, samples of preambles of incoming packets of an incoming packet transmission signal from a wireless device; outputting, by the classifier, a classification indicative of one or more coding parameters of the incoming packet transmission signal; transmitting a configuration command to configure the radio to receive the remainder of the incoming packet transmission signal in accordance with the one or more coding parameters indicated by the classification; Including, The method, wherein the one or more coding parameters are not pre-negotiated between the wireless device and a processing circuit executing the classifier prior to receiving the incoming packet transmission signal.

13. detecting a transmission rate from a portion of the preamble of the incoming packet transmission signal; adapting a radio configuration to receive the remainder of the incoming packet transmission signal at said transmission rate; The method of claim 12 further comprising:

14. 13. The method of claim 12, further comprising implementing an adaptive sampling algorithm for collecting the samples of the preamble of the incoming packet transmission signal.

15. In the wireless device, setting the encoding parameter to a value selected at the wireless device from among a plurality of pre-defined values ​​of the encoding parameter; transmitting the incoming packet transmission signal in accordance with the coding parameters without engaging in prior communication to pre-negotiate the coding parameters; The method of claim 14 further comprising:

16. filtering the incoming packet transmission signal using one or more bandpass filters, thereby generating a plurality of filtered incoming packet transmission signal components; determining that the signal is sufficient to determine the one or more coding parameters for one of the filtered incoming packet transmission signal components; The method of claim 14 further comprising:

17. The method of claim 12 , wherein the classifier is an artificial intelligence model including at least one convolutional neural network.

18. The artificial intelligence model is a multi-stage model, a first stage, wherein a bandwidth classifier includes a first convolutional neural network that classifies the incoming packet transmission signal into one of a plurality of bandwidth range classifications; a second stage in which, for signals having a bandwidth below a predetermined threshold, the signals are classified into one of a plurality of low-bandwidth coding classes by a low-bandwidth coding classifier comprising a second convolutional neural network, and, for signals having a bandwidth above the predetermined threshold, the signals are classified into one of a plurality of high-bandwidth coding classes by a high-bandwidth coding classifier comprising a third convolutional neural network; 18. The method of claim 17, comprising:

19. The method of claim 12 , wherein the one or more coding parameters include a bandwidth and / or a spreading factor.

Images