Patent Application
20200000366
Deepak Vasisht, Guo Zhang, Omid Abari, Hsiao-Ming Lu, Jacob Flanz, and Dina Katabi. “In-body backscatter communication and localization.” In Proceedings of the 2018 Conference of the ACM Special Interest Group on Data Communication (SIGCOM), pp. 132-146. ACM, August, 2018.
This application related to communication and/or localization using radio frequency backscatter from an in-body device.
The medical industry is looking at a wide array of in-body devices that include pacemakers that communicate their data over the wireless channel, smart pills that image the gastrointestinal tract, and microscale robots that access organs through the bloodstream. Today, such deep tissue systems communicate by generating their own radio signal, a process that consumes a lot of energy. For instance, in wireless capsule endoscopes, RF consumes 4 to 10 times more power than the sensors [35]. As a result, these capsules use large batteries that occupy about 40-50% of the space of the capsule [3, 6]. Reducing the power requirement for RF transmissions can reduce the size of the capsules making them more easy to swallow. It can also improve completion likelihood. Past work has found that 16.5% of the times, capsule endoscopes fail to completely visualize the small bowel primarily due to limited battery life [26].
Similarly, interest in deep tissue localization is on the rise. Localization of deep tissue sensors like capsule endoscope can enable physicians to isolate the parts of the GI tract with abnormalities, adapt video frame rates based on location, and deposit biomarkers at specific locations [27, 31]. The localization requirements for such capsules are on the order of a few centimeters [27].
Past work has tried to tackle these problems along multiple axes. Researchers have considered wireless power transfer—i.e., charging an implant using RF signals [21, 1]. These systems typically operate in the midfield where the RF transmitter is either directly in touch with the body or within a few centimeters from it.
The literature also has few proposals for in-body localization. One line of research uses magnetic field analysis [27, 2, 14]. The advantage of using the magnetic field is that its properties do not change much between air and human tissues. The disadvantage however is that the magnetic field power decays with a factor d6 as it travels through air [8]. Hence, the magnetic receiver (the receiving coil) has to be in touch with the body surface or within a few centimeters. Further, magnetic implants can be problematic. They can be painful if the person is exposed to a strong magnetic field as in the case of MRI [17]. They can also affect MRI images making it difficult to detect a tumor in the area near the implant [17]. A result, this form of localization is not widely used. Doctors also use X-ray or sonar for localization. These methods are expensive. Further, continuous tracking of an implant requires excessive x-ray exposure which increases cancer risk [9]. Finally, the use of ultrasound for in-body localization requires direct contact with the human skin, making it infeasible for several medical applications. For instance, presence of metallic equipment close to the human body can be an hindrance for administering X-ray/proton beams used for radiation therapy in cancer treatments [24, 10].
Some proposals for in-body RF-based localization have used the received signal strength (RSS) [34, 33]. Those systems use an array of receive antennas and either assume the implant to be closest to the receive antenna with the highest power or use path loss models to estimate location [31]. Analysis of the error bounds on RSS in-body localization has reported lower bounds of 4 to 6 cm [34] even when using up to 50 receive antennas. Past work has also tried to adapt indoor localization based on time-of-flight (ToF) or angle of arrival (AoA) for the domain [32, 22, 5]. Unfortunately, these systems are based purely on simulation, lack any empirical results, and most of them ignore signal deflection.
There is also a rich literature related to backscatter communication and localization in-air [4, 36, 15, 19]. However, in general, this work does not address problems that stem from RF propagation in deep tissues, such as signal deflection and body surface interference. The design in [36, 15, 18] proposes shifting the frequency of the backscatter signal to avoid WiFi interference. Also, some past work on harmonic RADARs and RFID-based localization ([28, 7, 13, 12]) has used non-linearity to mix two frequencies and weed out unwanted reflections from the environment.
In-body devices are becoming increasingly common in the medical industry. Some examples of in-body devices include pacemakers, smart pills (e.g., endoscopic pills), and microscale robots. Many in-body devices communicate with devices outside of the body by emitting radio frequency (RF) signals. For example, smart pills emit RF signals to communicate sensor data to receivers outside of a patient's body. To overcome the high attenuation associated with transmitting RF signals through multiple centimeters of tissue, many conventional in-body devices are active devices that transmit using battery power. One disadvantage of conventional, battery powered in-body devices is that their batteries often increase the size of the devices (e.g., the battery occupies more than 40% of the device).
For in-body devices that operate in deep tissue (i.e., centimeters below the skin), it is often the case that localization of those devices is required. For example, a gastroenterologist may need to know where an endoscopic pill is located in the gastrointestinal tract when a particular image is transmitted from the pill.
Aspects described herein are related to the use of backscatter communication techniques for communication with and/or localization of deep tissue in-body devices. Very generally, an external device is able to communicate with and localize a deep-tissue, passive backscatter device in the presence of skin reflections and through a number of layers of tissue, each being associated with a different propagation speed of radio frequency signals.
In some examples, the external device transmits RF signals at two or more different frequencies from one or more transmit antennas. The transmitted signals traverse one or more layers of tissue of a subject and reach an in-body device, causing the in-body device to emit a backscatter signal. The in-body device is a passive backscatter device that includes a non-linear circuit which causes the emitted backscatter signal to include a known mixture of the frequencies transmitted from the transmit antennas.
The backscatter signal and skin reflections of the transmitted signals are received at a number of receive antennas. Prior to digitization, the skin reflections (which are generally much stronger than the backscatter signal) are filtered out, leaving only the backscatter signals.
The backscatter signals received at the receive antennas are processed to localize the passive backscatter device in the subject's tissue. The processing accounts for the effects of different propagation speeds of radio frequency signals in the one or more layers of tissue.
In some examples, the internal device (e.g., “implant”, “smart pill”) has the ability to harvest power from received radio-frequency signals, while leveraging backscatter to communicate at zero power and save its harvested energy for its sensing tasks.
In some examples, the approach separates skin reflections from the signal from in-body implants, and solves unique localization challenges (like signal deflections, change of wavelength in-body) that do not exist in in-air localization.
In one exemplary application, aspects are used in cancer radiation therapy for accurate localization of tumors. For example, if, during cancer radiation treatment, the tumor being treated moves, there is a potential for significantly damaging tissues around the tumor. This is particularly a problem with proton therapy where the main benefit is the ability to accurately radiate at a particular depth with minimal exposure to the surrounding tissues. However, the tumor could move during the process causing the radiation beam to fall on the wrong tissue. For example, a breast cancer tumor may move with a patient's breathing, or a prostate cancer tumor may move due to bowel movements. Aspects described herein use wireless radio signals to track tumor movements during radiation by embedding a small marker (e.g., a passive backscatter device) in the body to assist in pinpoint and tracking tumor locations. In some examples, if localization of the passive backscatter device indicates that the tumor under treatment has moved beyond some threshold, the radiation stops until the tumor returns to its prior location. Alternatively, the beam could be steered to track the change in tumor position.
In a general aspect, a localization method includes receiving, at each antenna of a number of antennas, an emitted signal from a passive device located in a subject's body, each antenna of the number of antennas providing a respective received signal of a number of received signals, wherein the emitted signal includes a first set of frequency components and is caused by subjecting the passive device to a transmitted signal including a second set of frequency components not included in the first set of frequency components and processing the number of received signals to determine a location of the passive device, the processing being based at least in part on the effects of different propagation speeds of the emitted signal in one or more layers of tissue through which the emitted signal passes to reach the number of antennas.
Aspects may include one or more of the following features.
The passive device may include circuitry for forming the first set of frequency components from the second set of frequency components. The circuitry may include non-linear circuitry. The circuitry may include a diode. The transmitted signal may include a first frequency component with a first frequency and a second frequency component with a second frequency. The first set of frequency components includes a mixture of the first frequency and the second frequency. The method may include processing the received signals to remove components at frequencies of the second set of frequency components. Determining the location of the passive device may include determining a first set of distances between the number of antennas and the passive device and determining a second set of distances by processing the first set of distances according to the effects of the different propagation speeds of radio frequency signals in the one or more layers of tissue through which the emitted signal passes to reach the number of antennas.
The effects of the different propagation speeds of radio frequency signals in the one or more layers of tissue through which the emitted signal passes to reach the number of antennas may include refraction and changes in wavelength. The location of the passive device may be determined, at least in part, using a model of human tissue. The model may define the one or more layers of tissue as including an oil-based tissue layer and a water-based tissue layer.
In another aspect, in general, an in-body device includes an antenna and a non-linear circuit coupled to the antenna. The combination of the antenna and the non-linear circuit is configured to, when excited by a radio frequency signal including signal components at a first set of two of more frequencies, emit a radio frequency signal including signal components at a second set of frequencies that is distinct from the first set of frequencies.
Aspect can include one or more of the following features.
The non-linear circuit comprises a diode. For example, the non-linear circuit comprises a Schottky detector diode. The non-linear circuit may be a passive circuit.
The device further comprises circuitry configured to modulate the emitted radio frequency signal.
The circuitry configured to module the emitted radio frequency signal comprises a modulating element coupled to the antenna and the non-linear circuit. For example, the modulating element comprises a transistor.
The circuitry configured to modulate the emitted radio frequency signal comprises transmission circuitry for receiving data and outputting a control signal from controlling the modulating element.
The device comprises a sensor configured to acquire sensor data in the body and wherein the device is configured to modulate the emitted radio frequency signal according to the acquired sensor data. For example, the sensor comprises at least one of a camera, and electrical sensor, and a biochemical sensor.
The device comprising an energy harvesting component coupled to the antenna configured to convert received radio-frequency energy to power for operating circuitry of the in-body device.
In another general aspect, a kit includes an external device configured to perform any or all of the steps described above. Aspects may include a passive device for introduction into a subject, the passive device configured to perform any or all of the steps described above.
Overview
Referring to
In some embodiments, two patch antennas 102a may be used for transmissions and three patch antennas 102b (not all shown in
As discussed in more detail below, the transceiver emits radio signals at two (or more) different frequencies, and the backscatter device causes a combination of the signals at those two frequencies to be combined in a non-linear manner causing the signals emitted from the backscatter device to include components at other frequences than the two frequencies emitted from the transceiver, generally at harmonic combinations of those emitted frequencies. To avoid mixing of the transmitted frequencies in the transmission circuit causing harmonic combinations of the two frequencies, separate transmit chains may be used for each of the transmitted frequencies. One choice for the transmit frequencies is 830 MHz (f1) and 870 MHz (f2), with two harmonics received from the backscatter device being at 910 MHz (2f2−f1) and 1700 MHz (f1+f2). Of course other transmitted frequencies and harmonics may be used. In at least some embodiments, instead of transmitting a constant frequency signal while localizing or communicating with the backscatter device, the transmitted signals sweep through 8 MHz of bandwidth. As discussed further below, such sweeping may be used to improve localization accuracy.
In some embodiments, the antennas may be placed from 50 cm to 2 m away from the subject, and may be connected to Universal Software Radio Peripheral (USRP) software radios. The backscatter module 106 is located within a body 110 (e.g., surgically emplanted, introduced into the digestive tract, etc.). The body is permeable with respect to the signals produced by the transmitting antennas 102a of the transceiver 102. In some configurations, the body 110 is placed on a surface 112. In some embodiments, the setup can be placed below or above a bed or on the side, relative to body 110. The in-body module may be a small unit that can be attached to standard in-body devices that need to communicate data or be localized. While it is common to perform medical procedures while the patient is lying on a bed, the operation does not necessarily require the body 110 to be in a particular position.
Further referring to
As discussed further below, one embodiment of the backscatter module 106 used a diode connected to an antenna of the module. In operation, given a received signal from the multiple transmit antennas 104a (i.e., the received signal having multiple frequency components), the diode present in the backscatter module 106 causes the mixing of the frequencies of the components in the received signal thereby creating second and third order harmonic frequencies. In order to communicate from the backscatter device, a switch modulates (e.g., on-off) the connection of the antenna and the diode, thereby modulating the backscattered signals. The transceiver then demodulates the backscatter signal using corresponding demodulation techniques.
Referring to
Further referring to
In some embodiments, the controller 204 provides an input signal to the Localizer 214 that includes information to parameterize the collected location data. In further embodiments, this input signal from the controller 204 to the localizer 214 contains information about the signal created by the signal generator 206, which is compared against the measured result supplied by the filter 210.
Referring to
In general, the body 110 may be considered to be made up of multiple layers of body material, each with different radio propagation characteristics. In
Further referring to
Referring to
Further referring to
Referring to
In operation, in-air effective distance does not translate into physical distances directly. Furthermore, since the signals do not travel in a straight line or at constant speed, distance cannot simply be intersected from different transmitters and receivers to get an accurate location. In some embodiments, signal propagation is modeled as linear splines (instead of a straight line). This allows for propagation in each layer to be represented as linear while allowing for a change of direction across layers. In further embodiments, modeling the individual segments of the splines as functions of the latent variables in the model facilitates leveraging the observed effective in-air distances to estimate the latent variables. By doing so, this optimization can accurately estimate the position of the device by modeling the spline structure.
Further referring
Principles of Operation
Principles of operation of one or more embodiments are presented below, without necessarily limited the approach to be based or or to be consistent with these principles.
RF Signals in the Body
The manner in which RF signals behave as they propagate in biomaterial (e.g., fat, muscles) have implications for in-body backscatter, and therefore in the details of the design of the localization aspects of the system. From the perspective of electromagnetic (EM) waves, each material is characterized by two parameters: relative electrical permittivity, ϵr and relative magnetic permeability, μr. These are complex numbers that capture how the electrical and magnetic fields in an EM wave interact with the material. Both ϵr and μr are 1 for air and vacuum. For biological tissues, the relative magnetic permeability μr can be approximated as 1, so we set μr=1 in embodiments of the localization approach. However, ϵr has high variability depending on the tissue type and frequency of transmission. For example, for frequencies around 1 GHz (commonly used by in-body implants), the value of ϵr in muscle is 55−18j [16].
The value of ϵr is very important because it changes the speed of light and other electromagnetic waves (EM) in a material. Specifically, the speed of light in a biomaterial (e.g., muscle, fat, skin) is given by:
where c is the speed of light in vacuum and (to a good approximation) air. The change in the speed of the EM wave has important implications.
For a signal at frequency f, traveling in free space from a transmitter to a receiver separated by distance d, the wireless channel h(f,d) is given by
where A is the attenuation constant that depends on the antenna beam patterns and c is the speed of light in vacuum.
For biomaterial, incorporating EM wave speed change in equation (1) gives us the wireless channel, hM(f,d):
To understand the impact of ϵr on wave propagation, let us write √{square root over (ϵr)}=α−βj, where α and β are positive real numbers. The channel equation can then be updated as:
Note that the term
causes exponential loss in magnitude of the signal during propagation. The higher the value of β, the higher the loss. This is in addition to the propagation attenuation experienced by the signal in free space, given by
Also, muscle tissues experience significant additional loss in comparison with in-air signals.
Based on the considerations outlined above, in-body RF signals should use relatively low frequencies to avoid the drastic power loss occurring at higher frequencies. In fact, it is a common practice to use frequencies about 1 GHz, which are small enough to have a relatively low loss, but also large enough to enable relatively small electronics and antennas. Also, for backscatter signals which have to traverse the body twice, they lose more than 20 dB just to get 5 cm deep.
The electrical permittivity of a material further affects the efficiency of in-body antennas. As an antenna is placed in-body, its radiation efficiency decreases and its inherent losses increase as a function of ϵr. For muscle tissues, these effects incur another 10-20 dB of loss depending on the antenna design.
Consider again Eq. 3. Note that the signal phase changes much faster in biomaterial than in air. Specifically, the phase changes a times faster in biomaterial than in air. This is because the wavelength is a times smaller. This property is useful for RF-based localization algorithms that leverage phase changes to measure distance because it increases sensitivity and allows for measuring smaller distances (for the same signal SNR).
The value of ϵr affects not only how the signal travels through a material, but also affects what happens at the interface between two materials. Consider a signal traveling from a material with relative permittivity ϵr1 to a material with relative permittivity ϵr2. Further, for ease of exposition, assume that the signal is traveling perpendicular to the interface, which is the direction of minimum reflection. In this case, the ratio of the reflected power Pr and incident power Pt is given by:
As can be seen in the equation, larger the difference between the properties of two materials, the more signal power is reflected.
As discussed above, when an RF signal traverses the interface between two materials, it experiences a change in direction. This bending in the signal is called refraction. The relationship of the angle of incidence (θi) and the angle of refraction (θt) can be approximated by the following equation (for exact equation, see [25]):
Re(√{square root over (ϵr1)})sin θi=Re(√{square root over (ϵr2)})sin θt (5)
where Re(.) denotes the real part of a complex number. The RF signals therefore experience significant bending at skin-fat, fat-muscle and air-skin interfaces. This is the key challenge for localization of a device implanted inside the body. Based on this refraction, this means that it does not matter how the signal arrives in air, it enters the body almost along the direction of the normal on the surface. Since EM wave paths are reversible, this result means that all RF waves that exit the body must arrive at the skin-air interface almost perpendicular to the body surface. RF waves that try to exit the body with an angle far from the normal are not allowed to do so, and are reflected back inside the body. This observation is used in the localization approach.
Principles of Positioning
Conventional state-of-the-art RF localization systems generally operate in two steps. In the first step, they use the phase of the wireless channel between the transmitter and the receiver to measure the angle-of-arrival of the signal or distance between the transmitter and the receiver. In the second step, they assume the path traveled by the signal is straight, and apply basic geometry to locate the transmitter. For in-body RF signals, both these steps are bound to fail if applied as is.
First, let us consider the phase of the wireless channel. As shown in Eq. 1, the phase in air or vacuum, ϕ, is given by:
mod 2π. Here, j is the frequency of the signal and c is the speed of EM waves in vacuum. Hence, the phase of the wireless channel linearly depends on the distance traveled by the signal. In contrast, the phase accumulated by the signal in a biomaterial is scaled by α=Re(√{square root over (ϵr)}), where ϵr is the electrical permittivity of the material. Thus, when the signal traverses the body, its phase is:
where the sum is over the various materials traversed by the signal (air, skin, fat, muscle, etc.), αi is the real square-root of the electrical permittivity of material i, and di is the distance traveled in that material. This means that when the signal traverses multiple materials, the phase is no longer a simple function of the distance between the transmitter and the receiver.
Second, even if we could measure the distance traversed or angle-of-arrival of the signal, how does one map it to a location? The signal experiences refraction (bending) at the interface between different material (e.g., the interface between air and skin). Thus, the assumption that the signal travels between two points along straight lines no longer holds. As a result, the geometric model of intersecting distances or angles from multiple viewpoints doesn't apply.
Third, we need to take into account that our signal travels the entire distance at different frequencies. The signal from the transmitter to the backscatter device consists of two frequencies (f1 and f2), while the signal measured at the receiver after being backscattered is at f1+f2, 2f1+f2, and other frequency combinations. Together, these three factors are the main challenges facing in-body localization.
Before introducing a specific positioning algorithm used in one or more embodiments, we discuss a few key insights on which the algorithm is built. The first insight is that RF signals exit the body from a small region on the surface. Above, we made the observation that it does not matter how the signal arrives from air, it enters the body only close to the direction of the normal on the surface. Since RF propagation is reversible, this means that it is also not possible for the RF signal to escape from the body through all possible directions. In fact, it can escape only from a small region around the normal on the surface, as shown in
As a corollary of the first point, in-body multipath either does not exist or is very weak compared to the direct path. Any signal that is reflected back into the body has to traverse multiple cm of human tissue and face multiple reflections before it can escape the human body. Because of the exponential attenuation caused by human tissue, this signal will be very low power compared to the direct path emanating out of the body. This is quite in contrast to large scale in-air localization systems where the line-of-sight path can be much weaker than multipath because of obstructions.
Finally, the human body has multiple layers of tissues interleaved with each other. For example, skin and muscle are alike in electrical properties but are separated by fat which is closer to air.
Further, the same material can appear in multiple layers (e.g., air-skin-fat-muscle-fat-msucle). This complex layering structure makes it challenging to model refraction at various interfaces. However, for parallel layers, order and interleaving can be changed with no impact on the total phase of the signal. (Note that reordering of layers does affect the amplitude due to more reflections.) Since human tissues tend to be layered on top of each other, the assumption of parallelism is a reasonable approximation. This observation implies that the multiple layers of the human body can be rearranged for modeling and approximated to be grouped in two major layers: one layer comprising oil based tissues (like fat) and another layer comprising water based tissues (like skin and muscle).
Positioning Algorithm
A particular localization algorithm used in one or more embodiments has two steps. First, it estimates the distances traveled by the signal as if it were traveling in air. We call such values the effective-in-air distances. Second, it models signal paths with linear splines (piecewise linear segments). The length of each segment refers to the stretch of the path in a particular material (air, fat, muscles). It then solves an optimization problem that maps the effective distances to the correct splines that match the actual paths traveled by the signal. (For simplicity, all phase equations are expressed ignoring the initial difference in oscillator phase between transmitter and receiver which can be measured during the calibration phase.)
Measuring Effective in-Air Distances
Consider a signal traveling from a transmitter to a receiver through L different biomaterials. Assume that it travels distance di in biomaterial i, with phase scaling factor αi=Re(√{square root over (ϵri)}). We define effective in-air distance, deff, traveled by the signal as:
Combining Eq. 7 with Eq. 6, the phase, ϕ, of the signal observed by the receiver is:
Thus, an alternative definition for the effective in air distance is that, if traveled in air, it would result in the received phase.
So, how do we compute the effective distances? Recall that the system has two transmit antennas that transmit f1 and f2, respectively, and a number of receive antennas. Let d1 and d2 be the effective distances from the two transmitters to the backscatter device, and dr the effective distance from the backscatter device to receiver r. The transmitters are transmitting frequencies f1 and f2, while the receivers receive the non-linear mixing of these two signals at frequencies f1+f2, 2f1−f2, and other linear combinations. Let us consider the phase of f1+f2 measured at receive antenna r, which can be given by:
This phase equation is a combination of three components. The first two components correspond to the phase of the signal from the transmit antenna to the device. They combine based on the particular non-linear component of the signal that we receive. Since, we are considering just the non linear component f1+f2, which is just the sum of the frequencies, the corresponding phases also add up. For example, if we were to consider the frequency component 2f1−f2, then f1d1+f2d2 in Eq. 9 would be replaced by 2f1d1−f2d2. Eq. 9 gives us one equation in terms of three unknowns, d1, d2, and dr. We need more equations to solve for these unknowns. Note, now, that the non-linearity generates various frequency mixes, which provide additional equations. For example, we can write a similar equation for 2 f1−f2. The phase, ψi, measured at this frequency is given by:
Once again, note that the phase accumulated by the signal combines in the same way as the frequencies.
To simplify Eq. 9 and Eq. 10, we combine them as:
Thus, we get equations expressed as summed distances from each of the transmitters to the receivers. At this point, we cannot use more harmonics to solve for individual distances since they will just yield equations that are linearly dependent on these two harmonics. However, we can use another receiver r′ to get two additional equations that are functions of d1, d2 and dr′. Thus, given at least two receive antennas, these four equations can be solved to obtain d1, d2, dr and dr′. More antennas can be used to improve accuracy of distance estimates. We note that all phase equations are mod 2π. To resolve ambiguity due to the phase wrapping around, the system optionally, uses a small frequency band around each of the transmitted frequencies—i.e., instead of just transmitting f1 and f2, the system sweeps through its transmission in a small band of 10 MHz around f1 and f2.
Mapping Effective Distance to Actual Location
Now that we have the effective in-air distances between the in-body backscatter device and the transmit and receive antennas outside the body, we want to map those effective distances to the actual physical location of the backscatter device. At this stage we can drop the distinction between the transmit and receive antennas and treat all effective distances in the same way.
As we said before, the in-air effective distance does not translate into physical distances directly. Furthermore, since the signals do not travel in a straight line, the distances cannot simply be intersected from the different transmitters and receivers to get the right location. To solve this problem, we model signal propagation inside the human body as linear splines (instead of a straight line). Thus, propagation in each layer is linear, but across layers, it can change directions.
As discussed above, human tissues can be classified as either oil (like fat) or water based (like muscle). Furthermore, different layers can be rearranged such that the muscle-based tissues occur together and the fat-based tissues occur together. Thus, we model the human body as a two-layer system, as shown in has three latent variables (X, lm, lf). In addition to these latent variables, the model has fixed parameters, Θ: the position of each antenna, Xi for i=1, . . . , N, and the permittivity of biomaterials ϵrf and ϵrm. The observations made by the model are the effective distance measurements, di, from the implant to each of the antennas. Then, the goal of the model is to estimate the hidden variable (X, lm, lf) given a set of observations.
Next, how does the model constrain the structure of the splines for each path? Let us consider the effective in-air distance di measured at the ith antenna. In our model, the effective in-air distance is modeled by a spline comprised of 3 different segments: an in-air segment of length dai, in-fat segment of length dfi. and in-muscle segment, dmi. Together, when these physical distances are scaled by their respective scaling factors and summed together, they should yield the effective-in-air distance di. The estimation of the individual segments of these splines is governed by two sets of constraints:
Re(√{square root over (ϵra)})sin θai=Re(√{square root over (δrf)})sin θfi
Re(√{square root over (ϵrm)})sin θmi=Re(√{square root over (ϵrf)})sin θfi (12)
We now have a system with 6 variables (dai, dmi, dfi, θmi, θfi, θai) and 6 independent equations (Eq. 12 and 13). This is solvable numerically using ray tracing methods. Finally, dai, dfi and dmi, thus obtained are functions of the latent variables (X, lm, rf). Thus, we denote the length of the segments of the spline corresponding to antenna i in air, fat and muscle by dai(X, lm, lf), dfi(X, lm, lf), dmi(X, lm, lf) respectively.
Now that we modeled the individual segments of the splines as functions of the latent variables in the model, we want to leverage the observed effective in-air distances to estimate the latent variables. Specifically, for each antenna, we minimize the L2-norm of the observed effective in-air distances (di) and the distance obtained by the scaled sum of the spline segments (dai(.)+αfdfi(.)+αmdmi(.)). As before, α=Re(√{square root over (ϵr)}). Thus, combining distance measurements from multiple antennas, our optimization function can be written as:
where {circumflex over (X)}, {circumflex over (l)}m, {circumflex over (l)}f are the optimal values of the latent variables. This optimization problem is convex in each of the hidden variables (X, la, lf) for ϵr value ranges of human tissues. Furthermore, it has one local maximum. It can be framed as a standard convex optimization problem and solved using convex optimization techniques. By doing so, this optimization can accurately estimate the position of the device by modeling the spline structure.
Alternatives and Implementations
Referring to
The non-linear circuitry that causes generation of non-linear combinations of the input frequency components can comprise a diode 630. For example, a Schottky detector diode from Skyworks Solutions [29] is used. Other passive or potentially active non-linear components may be used to induce the non-linear behavior that causes emission of the harmonics of the input frequencies.
In embodiments that provide an output data channel from the backscatter module, a data storage and transmitter component 624 may include circuitry for serializing data stored in or accessible to the backscatter module. Such a serialized data stream may be used switch a transistor 622 so that the generation of the non-linear components is gated in time. In embodiments in which there is no permanent or long-term power source, an energy harvester 626 may be coupled to the antenna 610 to convert received RF energy to power useable by the data storage and transmission component. Not shown in
Regarding the transceiver, as introduced above, two transmit antennas 104a and corresponding transmit chains may be used to keep the transmit frequencies separate and avoiding generating harmonic components before emission in to the air. However, with suitable circuitry (e.g., power amplifier linearization etc.) a single transmit chain and transmit antenna may be sufficient. Also as introduced above, whereas multiple receive antennas 104b are useful for localization, if only data communication from the backscatter device is needed, then a single receive antenna may be sufficient. Even in the case of communication only, multiple receive antennas at different locations may nevertheless be useful, for example, to increase the overall signal-to-noise ratio in extracing the coded data in the backscattered signal. Furthermore, more than three receive antennas may be useful to increase the number of transmit-receive antenna pairs, thereby providing more constraints that may be used to determine the location and values of the various latent variables (e.g., layer thicknesses and properties, etc.). Various physical arrangements of the antennas may be used, for example, being arranged in an array placed close to the patient's body.
As introduced above, the backscatter module may be coupled to a variety of in-body devices, such as pacemakers, smart sensors that image the body (e.g., intestine) or measure physical and/or chemical properties of the body. The combined sensor and backscatter module may be relatively permanently affixed in the body (e.g., surgical implantation) or may be transient in the body (e.g., a “smart pill” that is ingested). In some examples, the in-body device is controlled by the received RF signals, for example, being triggered to sense the body when the signal is received and/or to receive control information encoded in the received RF signals. In many embodiments, no batter or long-term energy storage device is incorporated in the in-body device.
Embodiments of the transceiver (e.g., the out-of-body component of the system) may be implemented in software, in hardware, or a combination of software and hardware. Software components may include instructions stored on non-transitory machine-readable media, and these instructions cause processors to perform steps described above. For example, the localization algorithm may be implemented in a software-based module of the system. As another example, the radio transmission and reception may use a software-based radio. Some modules may use hardware (e.g., analog circuit) implementations. For example, the filtering of the originally transmitted frequencies may be performed in hardware component between the antenna and a software based radio such that the received signal is not digitized until after the interfering frequency components are filtered out. Some implementations may used hardware components that include special-purpose digital circuitry to implement modules of the system. For example, such hardware components may include application-specific integrated circuits (ASICs) or field-programmable gate arrays (FPGAs). Embodiments of the backscatter module include passive circuit components, and may further include software based components for data transmission and access to data storage elements in the device.
It is to be understood that the foregoing description is intended to illustrate and not to limit the scope defined by the appended claims. Other embodiments are within the scope of the following claims.
This application claims the benefit of U.S. Provisional Application No. 62/692,310, filed Jun. 29, 2018, which is incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| 62692310 | Jun 2018 | US |
