• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    Method for determining geometric information in network devices in the millimeter wave band. A method for determining geometric information of network devices in the millimeter waveband comprising measurements (401), by at least one of the devices, of signal strength and signal-to-noise ratio of a transmission received by another device on the network millimeter wave band; the step of estimating angular information (402, 403) of the received signal to generate (404) a set of reported particles (Pi) comprising the initial values of the state of each particle reported and entered into a modified particle filter; the modified particle filter that evolves (41) the set of reported particles (Pi) and past particles (Pp) to obtain a set of evolved particles (Pe) that, at the same time, evolves to obtain a set of subsequent particles (Ppost) delivered by the modified particle filter. Finally, the modified particle filter returns (44) as a result the final values of the geometric information of at least one device extracted from the set of subsequent particles (Ppost) delivered. (Machine-translation by Google Translate, not legally binding)
    公开号:ES2725773A1
    申请号:ES201830297
    申请日:2018-03-27
    公开日:2019-09-27
    发明作者:López Guillermo Bielsa；Beltran Joan Palacios；Paolo Casari；Jörg Carsten Widmer；Navarro Adrian Carlos Loch
    申请人:Fund Imdea Networks；
    IPC主号:
    专利说明:

    [0001]
    [0002] Method for determining geometric information in network devices in the millimeter wave band
    [0003]
    [0004] Field of the Invention
    [0005]
    [0006] The present invention has its application within the telecommunications sector. More specifically, it refers to millimeter wave networks and systems (millimeter wave) for wireless location of client devices in millimeter wave scenarios.
    [0007]
    [0008] The present invention relates to a method for determining geometric type information (angles, orientation and position properties) of the devices in the millimeter wave networks (for example, IEEE 802.11ad) using a modified particle filter (PF).
    [0009]
    [0010] Background of the invention
    [0011]
    [0012] Communications in the millimeter wave band (millimeter wave) impose strict requirements on the hardware. Very high frequency, bandwidth and data rate require high processing speed in transceivers. To ensure that it is feasible to produce hardware for end consumers, the design of commercial devices is often adapted to specific applications.
    [0013]
    [0014] Millimeter band devices generally use clusters (array) of antennas for transmission, resulting in the formation of directional radiation patterns that focus the emission of the signal towards certain azimuthal and elevation angles. Although the simplified theoretical models that are used, for example, in current standards assume ideal emission patterns that only focus on the desired directions and do not emit any radio power in other directions, commercial hardware destined for the market normally produces highly radiation patterns. irregular. For example, commercial access points (AP) conforming to the IEEE 802.11ad standard use low cost phase antenna sets that result in radiation patterns with extremely irregular shapes that often lack a clearly defined main lobe in a specific direction. , and they can even have two or more power lobes similar to the main one.
    [0015]
    [0016] In addition, the AP firmware uses approximate signal to noise ratio (SNR) values when selecting the best transmission sector. While this is sufficient for communications in scenarios with a limited number of client nodes, this strongly limits other cases of use of millimeter waves, such as high precision positioning.
    [0017] Even so, the large bandwidth available in the 60 GHz band, as well as the high number of elements in the phase antenna assemblies allows, in principle, to design high-precision positioning systems. The integration of these systems with standard protocols (for example, IEEE 802.11ad) is crucial for the deployment of the next generation of services based on the location of the client nodes, but it is also a challenge and limits the design options available. In addition, current 60 GHz hardware often uses quasi-omnidirectional patterns for reception, and therefore does not perform measurements to identify the best pattern in the millimeter wave receiver. This further limits the angular information that such commercial hardware can provide for positioning. Since commercial 60 GHz hardware only provides poorly approximate channel status information and uses very irregular transmission patterns due to its reduced cost design, it is very difficult to estimate the output angles (AoD) of the customers' signal , and even more difficult to translate AoD information into position and orientation estimates.
    [0018]
    [0019] In millimeter wave networks, client devices typically communicate using highly directional transmission (TX) as well as electronic beam formation (“beamforming”) in reception (RX) to compensate for propagation losses incurred by the signal of millimeter waves, and to reach sufficiently long communication distances. As a result, it is possible to take advantage of the beam-forming operations to estimate the output angles (AoD) of the millimeter wave signal of the transmitter and the arrival angles (AoA) of the same signal at the receiver. In some cases, the transmitter or receiver communicate using quasi-omnidirectional patterns instead of directional ones, which eliminates the need for beamforming on that client device and improves robustness at the expense of shorter communication ranges. In turn, this restricts the available angular information to only AoA (or AoD, respectively).
    [0020]
    [0021] Finally, triangulation algorithms require that at least three APs be within the customer communication scope to determine its location. Since signal blocking in millimeter wave networks is common, the presence of a sufficient number of APs cannot always be guaranteed. Therefore, positioning systems that are capable of coping with situations where information is lacking must be designed.
    [0022]
    [0023] In short, it is very desirable to develop a system that implements both communication and location functions in a commercial platform of low-cost millimeter waves, and therefore suitable for the final consumer.
    [0024]
    [0025] Summary of the invention
    [0026]
    [0027] The present invention solves the aforementioned problems and overcomes the limitations of the state of the art, providing a method for determining the geometric properties (angles, orientation and position) of a millimeter wave device, even in the presence of the irregular irregular radiation patterns typical of commercial hardware The present method works only with AoA, or only AoD, or both. In the context of the invention, AoA, AoD and their combination are jointly called angular information. Angular information is obtained by solving a linear programming (LP) problem that produces a dispersed angular spectrum. The orientation and position of the client device are estimated together through a modified particle filter (PF). The modification of the PF involves the injection of so-called "informed" particles, which represent possible reliable estimates of the user's position, and which are calculated by solving a problem of location by angle differences (ADoA).
    [0028]
    [0029] The described method applies to any millimeter wave (millimeter wave) scenario including IEEE 802.11ad networks with access points (AP) and client devices, mesh millimeter networks (mesh), ad networks "millimeter wave" hoc, as well as "millimeter wave" implementations of infrastructure composed of base stations, repeater terminals and mobile clients. As an example, but without limiting in any way the applicability of the current approach, the present invention applies to IEEE 802.11ad networks with APs and mobile clients, where it is assumed that APs are static and mobile clients must be located. As the angular information is collected in different devices, a feedback mechanism is used to deliver the information to the components of the method that require it, such as the modified PF. For example, the modified PF can be executed in APs, in which case the Client devices use feedback mechanisms to send angular information not directly detected by the AP.
    [0030]
    [0031] In addition to determining the angle of arrival of the millimeter wave signal, the proposed method determines the orientation and location of the client.
    [0032]
    [0033] The proposed method comprises two main blocks to be able to locate a customer despite the limitations of commercial devices given by its reduced cost: i) a formulation of a simple linear programming (LP) problem that allows the network to estimate the angle of arrival of the millimeter wave signal, so that said information is compatible with SNR values measured by APs or customers for different transmission pattern choices; and ii) a modified particle filter (PF) that obtains estimates of the possible location of the device while it is in motion. Most importantly, block i) is not based on any simplifying hypothesis, such as the design of custom transmission patterns, the “triangular” pattern shape, or the availability of phase information; additionally, block ii) is based on a mechanism of low complexity for updating particles in the PF, and in an informed way of injecting new particles that are more likely to be generated at the actual location of the client, thus accelerating the convergence of the PF and considerably improving its accuracy.
    [0034]
    [0035] The present invention is based on an LP formulation that produces a dispersed solution. This provides the necessary angular information. The number of estimated angles depends on the quality of the signal-to-noise ratio (SNR) and the directivity of the radiation diagrams of the consumer hardware. For example, in a network with APs and clients, the client carries out a training phase of transmission patterns (or sector scanning) by sequentially transmitting a message through each of its available patterns. All APs near the client measure the quality of the corresponding received signal. Specifically, the SNR is collected with which the access points receive the scanning messages of sectors sent by the client. Instead of relating each sector identifier to a specific angle, the transmission channel decomposes into a dispersed representation that relates the power to the angular information of each propagation path. Then the information of all APs and all patterns is merged to estimate the location of the client. This is done using linear programming (LP) along with Fourier analysis. In addition to avoid location errors due to blockages a modified PF is used to limit the error of location, even if a customer is in the coverage area of less than three APs.
    [0036]
    [0037] The modified PF addresses the state of a particle, which typically includes the position, speed and fitness of the particle. The particles in the modified PF evolve each time new angle measurements are received. The set of previous particles in the PF includes both later particles evolved during the previous PF steps and "informed" particles when available. In each iteration of the algorithm new informed particles are generated whose initial state is calculated from the measured angle information solving the corresponding ADoA problem. Therefore, the reported particles are more likely to be generated at the actual location of the customer, accelerating the convergence of the PF and substantially improving its accuracy. The fitness of the particles, which represents the probability that the position of the particle is correct given the measurements, employs the Fourier analysis applied to a smoothed version of the angular spectra determined through the initial solution of the LP problem. The suitability of the particles also takes into account the probability that the SNR received corresponds to the solution found.
    [0038]
    [0039] In the event that the AP positions are unknown, the APs perform a self-learning phase, where all the departure and arrival angles between each two APs that are within their communication range are determined by the APs themselves. Finally, a problem of minimization of mean square error to determine the positions of the APs is solved.
    [0040]
    [0041] The present invention only uses information that can be collected passively, such as SNR information from sector scans according to the 802.11ad standard. This information can also be quantified with low granularity.
    [0042]
    [0043] One aspect of the present invention relates to a method for determining the geometric information of the client devices in millimeter wave networks, comprising the following steps:
    [0044]
    [0045] - collect measurements, by at least one of the devices of a millimeter wave network, of the signal strength and of the signal-to-noise ratio (SNR) of a transmission received from another device in the millimeter wave network;
    [0046] - estimate the angular arrival information of the received signals, to generate a set of reported particles (P¡) that are introduced into a modified particle filter, the set P¡ comprising initial state values of each reported particle;
    [0047] - evolving, by means of the modified particle filter, the set of reported particles (P¡) and a set of passed particles (Pp) generated randomly and entered into the modified particle filter for initialization, to obtain a set of evolved particles (Pe);
    [0048] - evolving the set of evolved particles (Pe) to obtain a set of posterior particles (PPost) delivered by the modified particle filter;
    [0049] - deliver the final values of geometric information of at least one device obtained from the delivered set of posterior particles (PPost), the geometric information finally obtained is delivered by the modified particle filter as output values.
    [0050]
    [0051] The present invention has a number of advantages over the state of the art, which can be summarized as follows:
    [0052]
    [0053] - The present invention converts imperfect antenna patterns and low resolution SNR measurements into accurate angular information. This applies to the departure angle information, the arrival angle information, or both when available. The proposed method also works in case the beam formation is only used for packet transmission (or only for reception), and for the reception of packets (respectively, transmission) the devices use a quasi-omnidirectional transmission pattern. Therefore, the algorithm also works if only AoD or only AoA information is available;
    [0054] - It is not necessary to know any information about the location and initial orientation of the clients or about the APs in order to estimate the angular information;
    [0055] - The present invention can determine all angles, even between the APs themselves, through an LP formulation, based on the estimated angular information. The better the antenna patterns, the simpler the LP algorithms that can be used;
    [0056] - The proposed method does not imply any overload and any modification of the standard millimeter wave protocols, such as 802.11ad, since the method only collects information that is available at the access points as a byproduct of operations normally carried out. Therefore, the location of the device it can be achieved with zero additional communication costs (except for feedback of the location information to the entity that processes the data), thanks to the information collected by the 802.11ad access points during the transmission pattern training phase;
    [0057] - The proposed method works in real time, since it can update the location estimate dynamically every time a device scans the sector; - The proposed method operates in real scenarios and can be implemented in commercial APs that communicate in the 60 GHz band using antenna clusters with electronic pointing and a very limited SNR resolution.
    [0058]
    [0059] These and other advantages will be apparent in light of the detailed description of the invention.
    [0060]
    [0061] Description of the figures
    [0062]
    [0063] In order to aid in the understanding of the features of the invention, in accordance with a preferred practical embodiment thereof and in order to complement this description, the following figures are attached as an integral part thereof, which has a illustrative and non-limiting nature:
    [0064]
    [0065] Figure 1 shows a millimeter wave reference scenario for the location of a customer.
    [0066]
    [0067] Figure 2 shows three diagrams corresponding to three different channel decompositions, using respectively MMSE, linear programming and linear programming with regularized patterns, in four access points.
    [0068]
    [0069] Figure 3 shows a regression model of power loss with distance, for the calculation of the parameters of the model used to calculate the likelihood of SNR measurements.
    [0070]
    [0071] Figure 4 shows a flow chart of a method for determining customer geometric information in millimeter wave scenarios based on particle filters, in accordance with a preferred embodiment of the invention.
    [0072] Preferred Embodiment of the Invention
    [0073]
    [0074] The issues defined in this detailed description are provided to assist in a global understanding of the invention. Accordingly, those skilled in the art will recognize that variations, changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. In addition, the description of well-known functions and elements are omitted for clarity and conciseness.
    [0075]
    [0076] The embodiments of the invention can be implemented in a variety of platforms, operating systems and servers, devices or applications. Any specific arrangement, implementation, or architecture presented in this document is provided for illustrative and understanding purposes only, and is not intended to limit aspects of the invention. For a clearer exposition, which also does not limit the aspects of the invention in any way, an 802.11ad network is considered, where customers use beamforming in transmission, the APs receive the signal in quasi-omnidirectional mode and the APs obtain information about customer beam training to estimate AoD information.
    [0077]
    [0078] The proposed method for determining the geometric information of the client devices in millimeter wave networks comprises: i) estimating the angular information of the millimeter wave signal of the client by means of a linear programming formulation and ii) estimating the location of the client in real time using a modified PF.
    [0079]
    [0080] A typical 60 GHz indoor network implementation with multiple APs per enclosed space is considered.
    [0081]
    [0082] i) Estimation of departure angles (AoD)
    [0083]
    [0084] From referring to the coordinates of the i-th AP i, and x coordinates of the client node to locate, as shown in Figure 1. In addition, 9 is named client node orientation with respect to an absolute coordinate system. For transmission, the client node can choose from a total of B transmission patterns, where each pattern b is defined by pb (0), for b = 1, ..., B, where 0 is the emission angle and pb (0) is the amplitude gain of pattern b towards 0.
    [0085]
    [0086] Each time a client performs pattern training, each AP records the received signal strength indicator, RSSI and the corresponding SNR for each pattern used by the client and detected by the AP. Then the access points send this Ab) information to a central server, where the location process is executed. The SNR, measured in dB by the ith AP, is called '* when the client node transmits with pattern b, and be pr! = lu "/ 2 ° the corresponding amplitude of the signal.
    [0087]
    [0088] The most critical problems for customer location include the lack of phase information (which transforms the problem into a non-linear problem, and prevents the use of the most typical algorithms of angular pattern decomposition), the approximate discretization of SNR values in a logarithmic scale (with a typical resolution of 0.25 dB) and the device firmware (which is often too slow to register RSSI and SNR information for all customer patterns). The last problem results in incomplete measurements.
    [0089] A typical assumption for angle estimation is that the energy transported by the line of sight (LoS) path from the client to any AP exceeds the energy of non-line of sight (NLoS) routes. To find the AoD in the path of LoS, ¿, which best corresponds to the SNR measures of the APs, it is sufficient to solve the following problem of minimum mean square error, MMSE:
    [0090]
    [0091] 0 i - arg min min »(*>)
    [0092] b £ eBi 0 apb ( 0)) '
    [0093] Equation 1 where Bi is the set of patterns for which the AP has been able to take measures of RSSI and SNR.
    [0094]
    [0095] Irregular patterns of a commercial millimeter wave device can generate NLoS path measurements (represented as dashed lines in Figure 1) with a power similar to LoS paths (represented as solid lines in Figure 1).
    [0096]
    [0097] Therefore, the hypothesis that the LoS path dominates RSSI and SNR measurements is not verified in practice, since the transmission patterns of millimeter wave devices can emit significant power through secondary lobes, resulting in a significant NLoS road energy that reaches APs. In Figure 1, two patterns of Talon AD7200 devices are shown as an example. It is considered that the client node is placed in the x position, and communicates with the AP in position a 3 through the pattern b 1, and specifically of its largest lobe directed towards the lower left corner of the room R. A secondary lobe just above the main amplify in the same way a way to reach NLoS 3, resulting in a not insignificant power NLoS way about the power of LoS path.
    [0098]
    [0099] To avoid the above hypothesis, it is assumed that the power received from the millimeter wave signals of the client node dominates the noise in the SNR measurements collected by the APs. Since neither the phase information of the AP nor the weights of the array of phases of the array are known, it is also assumed that the measured amplitude is less than the sum of the amplitude of the total paths, which represents the extreme case that All these paths interfere constructively. This makes it possible to formulate the problem of estimating AoD as the following LP, with variables “¿W:
    [0100]
    [0101]
    [0102]
    [0103] As an example, in the left and center panels of Figure 2, the results of the AoD estimation obtained by four different APs, AP1 , AP2 , AP3 and AP4 , are shown using the MMSE method of equation 1 and the LP method in Equation 2, respectively. The points in the outer circle of each subfigure correspond to real AoD. Figure 2 shows that the solution of the LP problem estimates several probable directions, among which at least one is in fact very close to the LoS path between the AP and the client. In contrast, the MMSE solution shows greater discrepancies between the estimated angle of the LoS path and the actual AoD of the same path.
    [0104]
    [0105] ii) Estimation of the position of the client node
    [0106]
    [0107] The variables estimated through the LP method of equation 2 can be use to estimate the position x and the orientation of the client node so that these geometric parameters are in agreement with the AoD estimated by the different APs. It is called
    [0108] angles * refer to the case where the customer is oriented according to a reference coordinate system, that is, cp = 0.
    [0109]
    [0110] An angular fitness function
    [0111] £ (x) = max Gx (ip)
    [0112] customer, like *, where and ¡is the delta distribution | X, ¥> _ / JX I n
    [0113] Dirac centered 1 Where - ui 'r and then
    [0114] ^; also ^ '’’) denotes the internal product, and <8> denotes the circular convolution in the domain of the angles.
    [0115]
    [0116] Gx (^) = ^ í í f (V) = ^ a j (éf'H
    [0117] As - ¡, using the Fourier transform properties,
    [0118]
    [0119] where ^ (*) and ^ 11 1 symbolize the discrete Fourier transform (DFT) and its Inverse, respectively, and the dependence on cp has been eliminated for clarity.
    [0120]
    [0121] The simplified formulation used above is accurate only if the AoD estimates are not affected by any errors. To compensate for measurement errors, variables n °) are replaced by a version processed by means of a convolution
    [0122] cyclic
    [0123] this form of smoothed LP obtained through the variables r '^.
    [0124]
    [0125] An objective function Á x) can be defined to measure the suitability of the estimated position for a client node against the power measurements made for the set of transmission patterns used by the client node. This function can be evaluated even if the angles involved do not match perfectly with the angles estimated by the APs. The function is defined as:
    [0126] £ (x) = m ax G x (<¿ )
    [0127] v equation 3
    [0128] It is emphasized that the formula in equation 3 can be calculated very quickly, since it only needs FFT / IFFT operations, products and the application of operators that calculate the maximum of a set of values:
    [0129]
    [0130]
    [0131]
    [0132] The objective function ^ (x) in equation 3 is used to develop a modified PF that processes the measurements collected by the APs and locates a client node of millimeter waves on the fly, in real time.
    [0133]
    [0134] A likelihood function D (x) based on distance is provided to determine the coherence of SNR measurements, given that the user is in a certain position x. To achieve this, it is considered that the maximum SNR, in dB, measured by the ith AP for the best transmission pattern of the client, follows a log-normal distribution, conditional on a path loss model, that is:
    [0135]
    [0136] where |. | is the Euclidean norm and 7jmax, (llx - a¿ll) is the expectation of the maximum SNR in
    [0137] according to the path loss model) dlx _ a * ID = K ~ 10 '/ l ° gio (llx - a * ID.
    [0138]
    [0139] The parameters k and q of the loss model can be estimated by measuring pairs of values of 7 * 3Ilx - a, |) and lOí / log10 (| x - a, |) in an empty room, for a set of positions and casual orientations the client node and the APs. Next, a linear regression is performed to estimate the parameters k and q. The result is shown in Figure 3. The likelihood function D (x) for the maximum measured SNR values can finally be written based on the distance between the client node and an AP i as
    [0140]
    [0141]
    [0142] As an example, in equation 4, the value of od = 15 dB can be set, which is large enough to cover further attenuations caused by a partial or total blockage of signal propagation. The distance probability function D (x) is used only for an approximate consistency check. In fact, the value generated by the D (x) function is not used to determine the location of the customer, but rather to help filter out spurious solutions when more than one location is possible from the AoD estimate.
    [0143] The proposed method to locate the client device in real time, while it is moving in the millimeter wave scenario, uses a modified PF, shown in the upper part of Figure 4. The upper part PF1 of the flowchart functions as a typical particle filter: particles are generated randomly for the initialization of the algorithm. These particles are sent to the algorithm, grouped into an initial set of particles, with the role of forming the first set of passed particles P p for the first iteration. The particles evolve 41 according to a mobility model. The evolved particles P e are grouped into a P Post set, and constitute the P p set for the next PF loops. The fitness of evolved particles P e se
    [0144] updated based on the objective function ^ x), and on the likelihood function D (x), calculated according to equations 3 and 4, respectively. These equations incorporate the measures collected by PAs.
    [0145]
    [0146] The particle generation procedure is modified to be incorporated into a single "informed" particle in the initial set of particles when the AP calculates a new estimate. In detail, after collecting and processing the angle measurements 401 through the LP 402 method, the "informed" particle is positioned at a location calculated 403 through an ADoA algorithm applied to the three angles estimated by the APs by which the variables to «(0) are maximum. This solution can be calculated in a closed form, since it is essentially an intersection of circumferences, which is obtained much faster than, for example, the result of x = argmaxx £ (x) | _a reported part | It has zero speed and an initial fitness value equal to one to meet the criteria for fitness normalization. When creating the set of posterior particles, a renormalization of the aptitude is applied to avoid that a too large weight is applied to the particles that evolve from the reported particle.
    [0147]
    [0148] As an initialization step, before any measurement, the total number of particles is randomized according to a Gaussian distribution whose mean and variance are equal to the mean and variance of the set of AP locations. These particles are passed to the algorithm as a set of past particles P p for the first iteration. At the beginning of each iteration 404 an informed particle P i is generated using an ADoA algorithm applied to the AoD measurements assuming they correspond to the LoS paths. It is emphasized that this could be impossible in some cases, eg, due to the lack of the three AoDs that are needed, or to measurement errors that prevent the calculation of a possible solution.
    [0149] If the reported particle P i can be calculated the modified PF derives 41 a certain number of particles evolved from the reported particle, and another determined number of particles evolved 41 from the set of passed particles, according to an evolution model of mobility On the contrary, all particles evolve from past particles.
    [0150]
    [0151] To evolve 41 a particle, first choose a random particle k. The procedure for evolving 41 an informed particle P i is identical, except that in this case all evolved particles have the same parent. The variables xk, vk and fk refer respectively to the position, velocity, and fitness of a particle k. The set of these three quantities constitutes the state of the particle. It is recalled that each particle represents an estimate of the position of a millimeter wave client node. A new particle evolves 41 through the following equations, which define a mobility evolution model:
    [0152] x '= Xfc VfcA t + A x iAv
    [0153] v '= Vfc A v,
    [0154] where the apostrophe 'indicates the state variables of the evolved particle.
    [0155]
    [0156] This mobility evolution model is a uniform movement model in a linear trajectory with uncertainty Ax ~ Af (o, o i) and Av ~ Af (o, <T2i) ^ respectively, in the xy position at the velocity v, and At is the time between the instant of the current measurement and the instant of the previous measurement. It is assumed that ox = 2 ov = 1 m / s, and the uncertainty Af on fitness is defined as Af = x '- (x *, vfeA t) = Ax AíAv so that Af ~ a ^ o. ^ to í2 ^)
    [0157]
    [0158] To calculate 42 the value of particle fitness, the objective function ^ (x) and the
    [0159] likelihood function D (x). Being
    [0160] Ax = Af ct2 / (ct2 Aí2 (t2)
    [0161] Av = Af A t2o2 / v ( p 2x A t2a 2)
    [0162] Finally, the final expression for updating the state of the particle k is:
    [0163] x '= Xfc Vfc A i Af
    [0164] v '= Vfc Av.
    [0165] To update the ability fk of the particle k based on measurements made by the APs, calculates
    [0166]
    [0167]
    [0168]
    [0169] The formula includes a multiplication by Á x) and D (x), which provides an aptitude update based on estimates of angles and SNR derived from the measurements.
    [0170] Finally, in the modified PF, the suitability value of the evolved particles of the past particles P p and of the reported particle P i are renormalized 43 by applying different weights.
    [0171]
    [0172] After each iteration, the set of particles subsequent P Post from the output of the modified PF 44 provides information about the position and velocity of the particle, and also the likelihood that this position is correct (fitness). The position and speed of each mobile client node k in the millimeter wave network is determined by the particle characterized by the highest aptitude.
    [0173]
    [0174] It is noted that in this text, the term "understand" and its derivations (such as "comprising", etc.) should not be understood in an express sense, that is, these terms should not be construed as excluding the possibility that what described and defined may include more elements, steps, etc.
    权利要求:
    Claims (10)
    [1]
    1. A method for determining geometric information of devices in millimeter wave networks
    - collecting measurements (401), for at least one of the devices of a millimeter wave network, of a signal strength and a signal to noise ratio of a transmission received by another device of the millimeter wave network;
    The method characterized by further understanding:
    - estimate angular information (402, 403) of the signals received by at least one of the devices, to generate (404) a set of informed particles (Pi) that is entered into a modified particle filter, where the set of reported particles (Pi) includes the initial value of the status of each reported particle;
    - evolve (41), by the modified particle filter, the set of reported particles (Pi) and a set of passed particles (Pp) that is generated randomly and entered into the modified particle filter for initialization, to obtain a set of evolved particles (Pe);
    - evolve the set of evolved particles (Pe) to obtain a set of posterior particles (PPost) introduced to the modified particle filter;
    - deliver (44), by the modified particle filter, as a result the final values of the geometric information of at least one device from the set of subsequent particles delivered (PPost).
    [2]
    2. The method according to any of the preceding claims, wherein estimating the angular information (402, 403) comprises the step of applying linear programming (402) to the set of measures collected (401) of the signal strength and signal to noise ratio to obtain angular information selected between angle of departure and angle of arrival.
    [3]
    3. The method according to any of the preceding claims, wherein estimating the angular information (402, 403) further comprises the step of positioning informed particles (Pi) by applying the method of difference of arrival angles (403) to the estimated angular information.
    [4]
    4. The method according to any of the preceding claims, further comprising the step of renormalizing (43) the set of evolved particles (Pe) by means of the modified particle filter, where the renormalization (43) is applied to the particles evolved from the reported particle (P) and past sets of particles (P p) and comprises giving different weights to the evolved particles from the reported particle (P i) and the evolved particles past particles ( P p ).
    [5]
    5. The method according to any of the preceding claims, wherein the state of each reported particle is selected from position, velocity, and fitness.
    [6]
    6. The method according to claim 5, further comprising the step of calculating (42), by the modified particle filter, a suitability value for each particle evolved in the set of evolved particles (P e ), using the state of Each particle informed.
    [7]
    7. The method according to claims 4 and 6, further comprising the step of renormalizing (43) the evolved value of fitness for each evolved particle (P e ).
    [8]
    8. The method according to each of claims 4-7, further comprising (44) the step of delivering the final position, speed and suitability values of at least one of the devices corresponding to the normalization of the position, speed and fitness of evolved particles (P e ).
    [9]
    9. The method according to all the preceding claims, wherein the at least one millimeter wave network device is an access point or a client node.
    [10]
    10. A product formed by a computer program comprising the code of a computer program capable of executing all stages of the method as defined in any of claims 1-9.
    类似技术:
    公开号 | 公开日 | 专利标题
    Sun et al.2015|Synthesizing omnidirectional antenna patterns, received power and path loss from directional antennas for 5G millimeter-wave communications
    Gigl et al.2007|Analysis of a UWB indoor positioning system based on received signal strength
    Va et al.2015|Basic relationship between channel coherence time and beamwidth in vehicular channels
    CN108351422B|2020-02-14|Positioning method, base station and mobile terminal in mobile network
    El-Sayed et al.2014|Evaluation of localization methods in millimeter-wave wireless systems
    US9629116B2|2017-04-18|Apparatus, system and method of estimating a location of a mobile device
    Bielsa et al.2018|Indoor localization using commercial off-the-shelf 60 GHz access points
    US8447330B2|2013-05-21|System for wireless location estimation using radio transceivers with polarization diversity
    Tomic et al.2015|Hybrid RSS-AoA technique for 3-D node localization in wireless sensor networks
    Dil et al.2010|RSS-based localization with different antenna orientations
    Mahapatra et al.2018|Localization based on RSSI exploiting gaussian and averaging filter in wireless sensor network
    Shen et al.2014|Estimating multiple target locations in multi-path environments
    Ahmadi et al.2016|Semantic wireless localization of WiFi terminals in smart buildings
    Ju et al.2020|3-D statistical indoor channel model for millimeter-wave and sub-terahertz bands
    Ho et al.2020|Decentralized adaptive indoor positioning protocol using Bluetooth Low Energy
    Bogdani et al.2016|Localization error modeling of hybrid fingerprint-based techniques for indoor ultra-wideband systems
    Gui et al.2020|Connectivity based DV-hop localization for Internet of Things
    ES2725773B2|2021-09-03|Method for determining geometric information in network devices in the millimeter wave band
    Saccardi et al.2016|Probe correction technique of arbitrary order for high accuracy spherical near field antenna measurements
    Cheng et al.2015|Localization of a moving non-cooperative RF target in NLOS environment using RSS and AOA measurements
    Bohidar et al.2015|A comparative view on received signal strength | based location estimation in WSN
    Üreten et al.2014|A reduced complexity iterative grid search for RSS-based emitter localization
    Koenig et al.2011|Multipath mitigation for indoor localization based on IEEE 802.11 time-of-flight measurements
    Coca et al.2013|Antenna radiation pattern influence on the localization accuracy in wireless sensor networks
    JP2021124488A|2021-08-30|Location determination based on phase differences
    同族专利:
    公开号 | 公开日
    US10819453B2|2020-10-27|
    US20200252148A1|2020-08-06|
    ES2725773B2|2021-09-03|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    CN103117964A|2013-01-09|2013-05-22|北京邮电大学|Method and device of detection of signal of 60GHz millimeter wave communication system|
    EP2911435B1|2014-02-24|2020-02-12|Alcatel Lucent|Apparatuses, methods and computer programs for determining information related to a path direction and for a transceiver of a communication device|
    JP6494492B2|2015-11-17|2019-04-03|パナソニック株式会社|Millimeter wave communication control method and millimeter wave communication control apparatus|
    US11197261B2|2016-08-12|2021-12-07|Nec Corporation|Service location method and system for mmWave cellular environments|
    US10548147B2|2018-01-31|2020-01-28|Hewlett Packard Enterprise Development Lp|Access point beam strength rankings|
    US10505619B2|2018-01-31|2019-12-10|Hewlett Packard Enterprise Development Lp|Selecting beams based on channel measurements|
    法律状态:
    2019-09-27| BA2A| Patent application published|Ref document number: 2725773 Country of ref document: ES Kind code of ref document: A1 Effective date: 20190927 |
    2020-01-30| FA2A| Application withdrawn|Effective date: 20200124 |
    2020-07-20| NE2A| Request for restoration|Effective date: 20200714 |
    2021-09-03| FG2A| Definitive protection|Ref document number: 2725773 Country of ref document: ES Kind code of ref document: B2 Effective date: 20210903 |
    优先权:
    申请号 | 申请日 | 专利标题
    ES201830297A|ES2725773B2|2018-03-27|2018-03-27|Method for determining geometric information in network devices in the millimeter wave band|ES201830297A| ES2725773B2|2018-03-27|2018-03-27|Method for determining geometric information in network devices in the millimeter wave band|
    US16/365,953| US10819453B2|2018-03-27|2019-03-27|Method for determining geometric information of devices in millimeter-wave networks|
    [返回顶部]