Logistics companies seek to increase supply chain efficiency through real-time tracking assets, and companies hope to increase productivity through monitoring employees and customers, so demand for positioning services is growing. While Bluetooth receive signal strength indicators (RSSI) can be used to estimate the distance between known fixed points, this technique is usually not accurate enough for indoor positioning systems (IPS) and asset tracking. However, the Bluetooth specification update allows positioning more accurate.
Specifically, the latest version of the Bluetooth Core Specification (V5.1) (called "Bluetooth 5.1" in the market) has increased the arrival of the angle (AOA) and the starting angle (AOD) function, so that developers easier The location of the Bluetooth transmitter is accurately determined in two or three dimensions.
This article is the first part of the two series of articles, which will introduce AOA and AOD to explain how the Bluetooth core specification allows us to make it easier to implement technology. This paper will then introduce a feasible platform for implementing the application.
RF direction technology
RF (RF) direction based on RSSI can be estimated by signal strength. Higher accuracy can be achieved by performing multiple distance measurements from different points. A large key advantage of RSSI is that each device only needs one antenna to avoid complexity, cost, and size of antenna array. The shortcomings of this technique are lack of accuracy, which is only 3 m - 5 m.
The second common directional technique is called "arrival time" (TOA), that is, the travel time of the radio signal from a single transmitter to a single remote receiver. This method also requires only one antenna for each device, but the deficiencies require that each device is equipped with high precision synchronization clocks. The positioning accuracy of the TOA system can be close to 1 m.
With the release of the Bluetooth 5.1 specification, the Bluetooth Technology Alliance (SIG) decides to support the third type of directional technology based on AOA and AOD.
Using the AOA method, the receiving device can track the arrival angle of the individual object, and the receiving device can use the angle and its position from the plurality of beacons and their positions (Fig. 1).
The reason why it decided to increase the functioning function in Bluetooth 5.1, in part because of some entrepreneurial companies, they have provided proprietary AOA and AOD solutions for low-power Bluetooth (BLE) products. With Bluetooth 5.1, developers are able to extract "IQ" signal data (in-phase and orthogonal phase information) from the BLE packet through the core specification, and easily utilize RF directional functions. This makes developers to make the location service application easier.
For example, the AOA method is suitable for tracking transceivers that transmit BLE. The transceiver sends a data packet that supports the direction of the direction using an antenna, and receives the packet by a multi-antenna "positioner". The positioner samples the IQ data from the signal packet, while switching between each active antenna in the array so that it can detect the difference between the signal, this phase difference is due to each antenna in the array and a single signal emission The distance between the antenna is caused. Then, the positioning engine uses phase difference information to determine the angle of the received signal, then determine the direction of the transmitter (Fig. 2).
If the calculation signal direction from two or more locators is combined, the emitter position can be determined (Fig. 3).
For the AOD method, the situation is just the opposite. In this scenario, the device with an antenna array is transmitted through each antenna. When each signal packet from the antenna from the array reaches a single antenna of the receiver, since the signal is different from the distance from the transmitter, there is a phase shift relative to the previous signal (Fig. 4).
The antenna of the receiving device extracts the IQ sample from the signal packet and forward it to the positioning engine, and then the positioning engine uses data to determine the angle of the received signal, then determine the direction of the transmitter. This system is suitable for indoor navigation and other applications, where the transmitter is a fixed reference point, and the receiver may be a consumer's smartphone.
Bluetooth 5.1 update
Bluetooth 5.1 requires the change to change the RF software protocol (or "stack"), and some hardware (radio) enhancements are required, depending on the chip manufacturer. First, a revised protocol adds a fixed frequency extension signal (CTE) to any Bluetooth packet for the direction. (The packet is not modified by other means, so it can be used for standard BLE communication.)
CTE is a modulated signal that plus a frequency of 250 kHz with a Bluetooth carrier frequency (add 500 kHz when using BLE, add 500 kHz) for 16 to 160 μs. This signal includes a "unbailed" sequence consisting of 1, and the transmission time is sufficiently long, allowing the receiver to extract IQ data without an interference effect on the modulation. Since the CTE signal is transmitted at the last transmit, the cyclic redundancy check (CRC) of the packet is not affected.
The second important new feature of the specification allows developers to simply configure protocols to perform IQ samples. This configuration includes setting sampling timing and antenna switching, which is critical to the accuracy of position estimation.
Although different IQ sampling timing configurations can be used, in general, one IQ sample is recorded every 1 or 2 μs in the reference period of each antenna, and the result is recorded in the random access memory (RAM) of BLE SOC. The phase of the received signal changes because it is sampled by different antennas in the array, as shown in FIG.
Recording IQ samples is just the first step in building a location service application. To accomplish this task, developers must design or select the best antenna array in the application, and must master the complex algorithms required to perform the directional calculation.
Calculate signal direction
Antenna array for tie is usually divided into three array types: uniform linear arrays (ULA), uniform rectangular array (URA), and uniform circular arrays (UCA). As the name suggests, the linear array is a one-dimensional, and the rectangular array and the circular array are two-dimensional. ULA is easier to design and implement, but its drawback is that the tracking device is always moved on the same plane to calculate the azimuth. Otherwise, accuracy will be affected. URA and UCA can reliably measure the azimuth and elevation angle (Figure 6).
The antenna array designed to be tied is not simple. For example, when the antenna is placed in an array, they will interfere with each other's response. In order to consider these effects, the estimation algorithm typically requires a predefined array response. For example, a popular commercial algorithm is formed by two identical sub-arrays in mathematical assumptions. Fortunately, for those who lack antenna expertise, commercial antenna arrays with specified features can be used.
Effective antenna arrays ensure that accurate IQ samples are collected. However, the raw data is not sufficient to determine the direction of the signal; these data must be processed, fully considering multi-path reception, signal polarization, and propagation delay, noise, and jitter.
Since RF is not a new discipline, there is already some mature mathematical methods, and it is possible to estimate the reachable angle based on IQ sample acquired in practical applications. The definition of the problem, that is, the estimated transmit (narrowband) signal arrives at the arrival angle of the reception array (the calculation of the starting angle) is very simple, but the mathematical calculation of the decisor is not so simple.
Basically, it is assumed that each antenna in the array has a set of IQ sample data, and the commercial algorithm first calculates the data vector "X" in accordance with the following formula (assuming the signal exists, and is a zoom sinusoidal narrowband signal) to calculate the data vector "x":
Among them "A" is the mathematical model of antenna array ("guide vector"),
"S" is an incoming signal, "N" is noise item.
Then use the X and the following formula to generate the IQ sample covariance matrix "RXX":
This sample coefficient difference matrix is then used as an input to the primary estimate algorithm. One of the most popular and authenticated algorithms used for frequency estimation and radio direction is a multi-signal classification (MUSIC). From a technical point of view, the MUSIC uses the characteristic vector decomposition of the covariance matrix and the emission value, based on the attributes of the signal and noise subspace, estimates AOA.
The formula used is:
"A" is a diagonal matrix containing the emission value, "V" is a matrix containing the corresponding feature vector.
Once V is isolated, it can be used in the formula to generate a tamper, and the peak appears in the arrival angle of the received signal (Equation 4):
The resulting spectrum is shown below, the peak appears in the direction of the transfer signal arrives (Fig. 7).
The amount of calculation of running the directional algorithm is large and requires enough RAM and flash capacity.
Commercial Bluetooth 5.1 products with corresponding resources have been listed. For example, Dialog Semiconductor offers DA14691 Bluetooth 5 Le So, which is suitable for positioning service applications. The chip uses the ARM® Cortex®-M33 microprocessor, providing 512 KB of RAM. Silicon Labs released the Bluetooth 5.1 stack of EFR32BG13 BLE SOC, which uses ARM Cortex-M4 microprocessors, providing 64 KB of RAM and 512 KB flash.
Nordic Semiconductor is further, and they released a new "direction" hardware NRF52811. This BLE SOC can be compatible with Bluetooth 5.1, integrated ARM Cortex M4 microprocessor, combined with multi-protocol radio from Nordic NRF52840 wireless SOC. The chip provides 192 kB of flash memory and 24 kb RAM.
Part 2 of this series will explain how to use these SOCs and stack development platforms (combined with other components, including antenna array, auxiliary microprocessor, and related memory, and "positioning engine" firmware) to achieve positioning service applications, For example, asset tracking and IPS.
Summarize
With the core specification enhancement function adopted by Bluetooth 5.1, users can more easily access IQ data. These data can be used to feed the RF directional algorithm to calculate the AOA or AOD transmitted by Bluetooth radio, and then utilize this information, the position of the transmitter is estimated in two or three dimensions.
Although these algorithms can lay the foundation for the actual positioning service applications such as asset tracking and IPS, their accuracy depends on the design reasonable antenna array and the verified RF directional algorithm, must have sufficient processor and memory resources to execute. Complex calculation.
In Part 2 of this series, we will discuss that although the development is not simple, as the commercial Bluetooth 5.1 measurement platform, antenna array and positioning engine firmware, designers can start building cm level accuracy easier. Location service application.
Our other product:
