Patent Application
20150295628
The present invention relates to a method for determining a time multiplexing sequence for a MIMO radar unit. The present invention also relates to a MIMO radar unit.
In motor vehicles, radar systems are increasingly used in order to determine distances and relative speeds of objects in the vicinity of the motor vehicle. The named radar systems can be used to realize convenience functions (such as adaptive cruise control, or ACC) and safety functions (such as warning the driver in critical situations, or causing a full braking of the vehicle if an impact can no longer be avoided).
For these applications, the distance and relative speed of the objects, and the angle of the reflected radar waves, have to be determined precisely. In particular, the estimation of the angle, or direction of arrival (DOA), has to be carried out precisely for this purpose. The angle to be estimated can be an azimuth angle and/or an elevation angle. Compared to conventional single-input multiple-output (SIMO) radar systems, multiple-input multiple-output (MIMO) radar systems offer the advantage of a larger virtual aperture at the same or smaller geometrical size. Therefore, normally a greater precision of the estimated DOA can be achieved using an MIMO radar unit.
In general, however, the precision of DOA estimations in an MIMO radar unit decreases if the object is moving relative to the radar unit, because the movement produces an unknown phase shift (Doppler phase) of the base band signal, due to the Doppler effect.
Feger et al., “A 77-GHz FMCW MIMO Radar based on a SiGe Single-Chip Transceiver” (2009) discloses a MIMO radar unit in which receive and transmit antennas are positioned such that at least two antenna elements of a virtual array have the same geometrical position (redundant position). An unknown phase shift (Doppler phase) of the base band signal due to a target movement is estimated by calculating the phase difference of the antenna elements at the redundant positions. The estimated Doppler phase is subsequently used as a correction in an algorithm for the angle estimation.
However, in order to determine the unknown Doppler phase, only antenna elements at the redundant positions are used. Subsequently, all antenna elements are used to determine the angle at which the signal arrives. Therefore, a noise of the antenna elements at the redundant positions has a greater influence on the DOA estimation than does a noise at the rest of the antenna elements. As a function of the time multiplexing schema used, and of the configuration of the antennas, a DOA estimation of such a system can even be worse than with the use of a single transmit antenna.
Wintermantel, “Radar System with Elevation Measuring Capability” (2010) discloses a MIMO radar system having a plurality of transmit and receive antennas. The transmit antennas transmit numerous chirp sequences of short duration, for example 512 or 1024 chirps each with a duration of, for example, 10 μs. The transmitters transmit sequentially or simultaneously, but with different phase modulations. The distance, the relative speed, and the DOA are estimated using a three-dimensional discrete Fourier transform. In particular, in the case of a plurality of objects having the same distance and the same relative speed, the proposed system can carry out only imprecise angle estimations, and in addition cannot be used at all with a frequency-modulated continuous wave (FMCW) radar having long ramps. In addition, due to the chirp sequence design with rapid frequency changes, the hardware is relatively complex and expensive to realize.
Therefore, an object of the present invention is to provide an improved method for determining a time multiplexing sequence for a MIMO radar unit.
According to a first aspect, the object is achieved by a method for determining a time multiplexing sequence for a MIMO radar that includes a plurality of transmit antennas. In the case where the plurality of transmit antennas includes more than two transmit antennas, the method is applied to the two of the plurality of transmit antennas whose distance from one another is the maximum of any other pair of the plurality of transmit antennas. The transmit energy of each pulse of the time multiplexing sequence transmitted by the two transmit antennas is essentially the same, a number and a transmit duration of the pulses are defined. Transmit times of the pulses are configured to be essentially equidistant, and the transmit times of the pulses are selected as follows, without limitation of generality: ti=i−(N+1)/2, where 1≦i≦N, and where the parameters are defined as follows:
The indices of the transmit times of transmit antenna TX1 are determined iteratively from the following mathematical equations:
I
LB
:=I
LB
, if ILB>imax
I
LB
:=i
max+1 otherwise
I
LB
=σ−σM-1+½·(1+(K−M)·(K−M−N)+N)
I
UB=(σ−σM-1)/(K+1−M)−(K−1−M−N)/2
where the parameters are defined as follows:
where σ and K are determined relative to N from the following table:
The transmit times of transmit antenna TX2 occur at the times at which transmit antenna TX1 is not transmitting.
Advantageously, in this way it is possible to determine the sequence in which the transmit antennas transmit. An angle estimation is advantageously possible with great precision using the determined sequences. As a result, using the method according to the present invention a transmit sequence is determined for the two transmit antennas that are used, the transmit sequence defining when which pulse is transmitted by which transmit antenna.
According to a second aspect, the object is achieved by a method for determining a time multiplexing sequence for a MIMO radar unit that includes a plurality of transmit antennas, where, according to the method, a number of pulses, transmit powers, and transmit durations of the transmit antennas are defined; in the case of more than two transmit antennas, the two transmit antennas whose distance from one another is the maximum of all pairs of the plurality of antennas are used; the pulses are divided between the transmit antennas in such a way that the magnitude of the difference between the transmit energy radiated via the first transmit antenna and the transmit energy radiated via the second transmit antenna is the minimum; a time multiplexing schema that provides a transmit sequence and transmit times of the transmit antennas is determined from the following mathematical equation [1 (1Tρ(1))·(ρ(1))T−1/(1Tρ(2))·(ρ(2))T]·t=0; and determining from the determined sequences, those that have the shortest overall transmit time, where the parameters of the equation are defined as follows:
In this way, it is advantageously possible to determine a time multiplexing sequence with a minimum transmit duration and that is optimized with regard to angle estimation.
In an example embodiment, in the case in which the number of pulses can be freely selected, a multiple of four is used for the number of pulses. In this way, the realization of the method according to the present invention is possible for an arbitrary number of pulses.
In an example embodiment, a deviation from the equidistance of the transmit times of the pulses is small in relation to the temporal spacing of the transmit times. In this way, it is advantageously supported that, even in the case of an imprecise equidistance of the transmit times of the pulse durations, the method according to the present invention can still be carried out.
In an example embodiment, the partitioning of the pulses to the transmit antennas is carried out by a solution algorithm of a partitioning problem. In this way, the allocation of the pulses to the transmit antennas can be carried out using a method that is well known in informatics.
In an example embodiment, the partitioning of the pulses to the transmit antennas is carried out by an approximation algorithm with a polynomial runtime. Under some circumstances, this algorithm does not always yield the best solution, but, on the other hand, it has a significantly shorter runtime, and is therefore advantageous in particular for larger numbers of pulses.
In the following, the present invention is described in detail with further features and advantages, on the basis of the figured. All described or represented features, in themselves or in arbitrary combination, form the subject matter of the present invention, independent of their summarization in the patent claims or their relations of dependence, and independent of their formulation or representation in the description or in the drawings. Known principles of a MIMO radar unit are not described in more detail.
In order to be able to make use of the advantages of MIMO radar, the transmit signals of the transmit antennas of the MIMO radar that are used should preferably be orthogonal to one another. To achieve this, in principle there are three possibilities, which however each have specific disadvantages: code multiplexing, in fast or slow time, has a high technical outlay, or has only limited orthogonality; frequency multiplexing causes a phase and Doppler shift, as a function of the wavelength; and time multiplexing results in different phases when there are object movements between changeovers, making the subsequent angle estimation more difficult. In the following, only the time multiplexing method is examined in more detail, because it is comparatively easy to realize in terms of circuitry, and is therefore economical. According to example embodiments of the present invention, a time multiplexing wiring design is provided that is distinguished in that object movements between changeovers of the transmitter do not have any influence on the subsequent angle estimation.
The received basis band radar signal is described by a mathematical model, the Cramer-Rao bound (CRB) being calculated for the angle estimation. In principle, this bound represents the maximum possible precision that can be achieved with an unbiased angle estimator, and is therefore independent of the evaluation algorithm that is used. From the Cramer-Rao bound, a condition is derived between the positions of transmit antennas TX1 . . . TXN, the transmit sequence, and the transmit times, that should be met in order to enable an angle estimation that is as precise as possible.
From this modeling, various optimal time multiplexing schemata can then be derived. In the angle estimation, not only the phase differences in the base band signals resulting from the angle, but also the phase differences resulting from the Doppler effect, are taken into account. The transmit times and positions of the respectively transmitting transmit antennas TX1 . . . TXN are selected such that the transmit positions of the transmit antennas TX1 . . . TXN are as uncorrelated as possible in the sequence in which they transmit and in their transmit times. In this way, for the angle estimation, a degree of precision is achieved that is exactly as great as when the target object does not move at all relative to the radar. Advantageously, this design can be used both for an elevation angle estimation and for an azimuth angle estimation. As a result, in this way a technical potential of the MIMO radar unit can advantageously be exploited as well as possible.
Without limitation of generality, for the ith transmit time, the following mathematical equation holds:
t
i
=i−(N+1)/2,1≦i≦N (1)
with the parameters:
I . . . index
ti . . . ith transmit time of the transmit antennas
N . . . number of pulses.
Thus, for example, for N=5, the corresponding transmit times would be given by: t1=−2, t2=−1, t3=0, t4=1, t5=2.
If the algorithm is in the Mth iteration, then a minimum index ILB (lower bound) and a maximum index IUB (upper bound) of the transmit times can be determined at which first transmit antenna TX1 is permitted to transmit as the next transmitting antenna. This means that all transmit times between tI
I
LB
:=I
LB
, if ILB>imax (2)
I
LB
:=i
max+1 otherwise (3)
where
I
LB
=σ−σM-1+½·(1+(K−M)·(K−M−N)+N)
(4)
I
UB=(σ−σM-1)/(K+1−M)−(K−1−M−N)/2
(5)
where the parameters are defined as follows:
where σ and K are determined relative to N from the following table:
All the time multiplexing sequences that are optimal for the angular estimation can be determined by solving equations (1) through (5). The algorithm iteratively determines the possible transmit times of first transmit antenna TX1. In this context, “iteratively” means that first all possible first transmit times of first transmit antenna TX1 are determined, for each first transmit time, in the subsequent iteration all possible second transmit times of first transmit antenna TX1 are then determined, etc. This is repeated until the corresponding number of transmit times has been determined. The transmit times of second transmit antenna TX2 thus result as those times at which first transmit antenna TX1 is not transmitting. These times therefore no longer have to be determined separately.
For better understanding,
Shown is a schematic sequence of the algorithm for a total of eight pulses P1 . . . P8 determined from four iterations IT1 through IT4. In first iteration IT1, all possible first transmit times of first transmit antenna TX1 are determined. It can be seen that after first iteration IT1, indices of possible transmit times of first transmit antenna TX1 are one, two, or three. As a result, this means that either the time one, or the time two, or the time three are possible first transmit times for first transmit antenna TX1.
For each of the iteratively determined first transmit times one, two, and three, subsequently, in a second iteration IT2, all possible second transmit times of first transmit antenna TX1 are determined, etc., until, after the fourth iteration IT4, all four transmit times of first transmit antenna TX1 have been determined At all other times, second transmit antenna TX2 transmits, this sequence resulting automatically because the transmit times of the transmit pulses of the two transmit antennas TX1, TX2 are essentially equidistant. The time multiplexing sequences shown as shaded are distinguished by a particular structure, according to which the two transmit antennas TX1, TX2 transmit equally often, and the sequence is made symmetrically.
After the termination of fourth iteration IT4, there are therefore eight possibilities for the definition of four transmit times of first transmit antenna TX1. In supplementation with the four transmit times of second transmit antenna TX2, there thus result eight possible time multiplexing sequences for the two transmit antennas TX1, TX2, all eight time multiplexing sequences being in principle equivalent.
In the following, two scenarios are distinguished for which the method according to the present invention can provide optimal time multiplexing sequences.
In a first scenario, each pulse P1 . . . PN has essentially the same transmit energy, the transmit times of the transmit durations of pulses P1 . . . PN being configured essentially equidistant from one another. A certain deviation from this equidistance is possible, which can result in a slightly worsened angle estimation. A number N of transmit pulses P1 . . . PN is specified, determined for example from an optimization of a distance and speed estimation of the MIMO radar unit.
For this scenario, as a function of the total transmitted pulses N, the following optimal time multiplexing sequences can be determined the number N of pulses P1 . . . Pn is a natural multiple of four, i.e., N=4, 8, 12, 16, . . . =4k, kε; only two transmit antennas, i.e., those two transmit antennas TX1, TX2, are used whose distance from one another is the maximum, standardly limited by an aperture size of the transmit antennas; and overall, an equal number of pulses are transmitted via the two transmit antennas TX1, TX2.
For the transmit times of first and second transmit antenna TX1 and TX2, in this case the following holds:
E
S(t(1))=ES(t(2)) (6)
with the parameters:
ES(x) . . . mean value of the elements of the vector x
t
(1) . . . transmit times of first transmit antenna TX1
t
(2) . . . transmit times of first transmit antenna TX2
A certain part of these optimal sequences has the structure that both transmit antennas TX1, TX2 are used equally often, and the time multiplexing sequence is symmetrical to the center of the sequence. This means that two pulses having the same distance from the center of the sequence have to have the same value.
The transmit schema of the time multiplexing sequence of
This sequence indicates that transmit antenna TX1 transmits at the first and second transmit times, transmit antenna TX2 transmits at the third, fourth, fifth, and sixth transmit time, etc. From the example, the symmetry of the sequence can also be seen, the center of the sequence being situated between the fourth and the fifth pulse.
Although in
In a case where the number N of the pulses P1 . . . PN has the form N=2, 6, 10, 14, . . . =2+4k, where kε, then all the optimal sequences have the following properties: only two transmit antennas, i.e., those two transmit antennas TX1, TX2, are used whose distance from one another is the maximum; overall, an equal number of pulses P1 . . . PN are sent via the two transmit antennas TX1, TX2; and for the transmit times of first and second transmit antenna TX1, TX2 the following holds:
|ES(t(1))=ES(t(2))| is a minimum (7)
A certain portion of these optimal sequences has the structure that the two transmit antennas TX1, TX2 are used equally often; and the time multiplexing sequence is symmetrical to the center of the sequence, except for those elements that are closest to the center of the sequence, which are different from one another. An example of such an optimal time multiplexing sequence having this specific structure is the following sequence: TX2, TX1, TX1, TX2, TX1, TX2, where a total of six pulses P1 . . . P6 are transmitted. Here, the two transmit antennas TX1, TX2 are used equally often, and the pulses situated closest to the sequence center, i.e., the pulses three and four, are the single pulses for which there is no symmetry.
Another optimal sequence that does not have this structure is the following sequence: TX2, TX1, TX1, TX2, TX2, TX1.
If the number N of pulses P1 . . . PN is odd, i.e., N=3, 5, 7, 9, . . . =2k+1, kε, then the following properties hold for all time multiplexing sequences that are optimal with regard to an angle estimation: only two transmit antennas, i.e., those two transmit antennas TX1, TX2, are used whose distance from one another is the maximum; via one of the transmit antennas TX1, TX2, one more pulse is radiated than is radiated via the other transmit antenna; and for the transmit times of first and second transmit antennas TX1, TX2, the following holds:
E
S(t(1))=ES(t(2)) (8)
A certain portion of these time multiplexing sequences that are optimal with regard to angle estimation has the following structure: one of the transmit antennas TX1, TX2 is used one time more than is the other transmit antenna; and the time multiplexing sequence is symmetrical to the center of the sequence.
An example of an optimal time multiplexing sequence having this specific structure is the following sequence: TX2, TX2, TX1, TX1, TX1, TX2, TX2, where a total of seven pulses P1 . . . P7 are transmitted. Here, transmit antenna TX2 is used one time more than is transmit antenna TX1. The sequence is symmetrical to element four, which corresponds to the center of the sequence.
Another time multiplexing sequence which is optimal with regard to angle estimation is the following sequence: TX2, TX1, TX1, TX2, TX2, TX2, TX1.
For a case in which the number of pulses N can be selected freely, the time multiplexing sequence that is optimal with regard to angle estimation is obtained if the number of pulses N is a multiple of four, i.e., N=4k, kε, and one of the time multiplexing sequences identified above as optimal is used.
In a second scenario, a case can occur in which the number of pulses N and the transmit powers and transmit durations of transmit antennas TX1, . . . TXN are already specified, because they were determined or optimized from a distance and speed estimation. In this case, it is sought to assign pulses P1 . . . PN to a plurality of transmit antennas TX1 . . . TXN in such a way that a best possible angle estimation is achieved using a time multiplexing sequence having a minimum overall transmit duration.
As in the scenario described above, for a best possible angle estimation, only two transmit antennas TX1, TX2, i.e., those two transmit antennas TX1, TX2, are used whose distance to one another is the maximum.
The individual pulses P1 . . . PN are divided between the two transmit antennas TX1, TX2, in such a way that the magnitude of the difference between the total energy radiated via first transmit antenna TX1 and the total energy radiated via second transmit antenna TX2 is a minimum. In the ideal case, an equal amount of transmit energy is radiated via the two antennas TX1, TX2. An allocation of transmit energies or pulses to transmit antennas TX1, TX2 does not have to be unique, so that there can exist a plurality of equally good allocations.
In a first step, the named optimal allocation is determined by a solution algorithm of the partitioning problem known from informatics. Because this is an NP-complete problem, the determination of an exact solution can be very time-intensive. However, for a small number of pulses N, the calculating time can be uncritical. For large numbers of pulses N, an approximation algorithm can be used having a polynomial runtime, which may indeed not yield the best solution, but advantageously has a significantly shorter runtime.
For each allocation of pulses P1 . . . PN to transmit antennas TX1 . . . TXN, in a following second step the possible transmit times are determined for which there result optimal angle estimations. The transmit times result from the solution quantity of a lower-order linear equation system, so that there are infinitely many transmit times that result in an optimal angle estimation. The transmit times t result from the following mathematical equation:
[1/(1Tρ(1))·(ρ(1))T−1/(1Tρ(2))·(ρ(2))T]·t=0 (9)
with the parameters:
Finally, from these transmit times, those transmit times are determined that minimize the overall transmit time, so that, as a result, a lowest possible overall duration of the time multiplexing sequence is determined. This is carried out for example by a linear program that minimizes the last transmit time point tN, subject to the condition that the difference between two successive transmit times ti and ti+1 must not be smaller than the duration of the ith pulse. In order to determine from this the transmit times for the minimum overall transmit duration, all permutations of the original solutions for transmit times t have to be checked. With this method, in the indicated scenario, the time multiplexing sequences for an optimal angle estimation with minimum overall transmit time can always be determined.
For simpler calculation, in a first step 200, the transmit times are selected without limitation of generality according to mathematical equation (1).
In a step 201, indices of transmit times of transmit antennas TX1, TX2 are determined from equation (1), the transmit times of first transmit antenna TX1 and a number K of pulses transmitted via first transmit antenna TX1 being determined iteratively from equations (2) through (5), and from Table 1.
In a first step 300, the pulses are divided between transmit antennas TX1, TX2 in such a way that the magnitude of the difference between the transmit energy radiated via first transmit antenna TX1 and the transmit energy radiated via second transmit antenna TX2 is minimized.
In a second step 301, transmit sequences and transmit times for transmit antennas TX1 and TX2 are determined from equation (9).
Finally, in a third step 302, from the determined sequences, those sequences that have the lowest overall transmit time are selected.
In sum, the present invention proposes an improved method for providing time multiplexing sequences for transmit antennas of a linear or two-dimensional MIMO radar unit that includes antennas situated close to one another (collocated MIMO radar), enabling an angle estimation that is as precise as possible.
Advantageously, the design according to the present invention is distinguished in that algorithms are proposed that differ in their precondition or boundary condition, with which a transmit sequence of transmitters and their transmit times can be determined, with which object movements have essentially no influence on the angle estimation. A switching design according to example embodiments of the present invention provides a time multiplexer switching of a defined number of transmitters in such a way that a maximally optimized angle estimation is enabled, according to the Cramer-Rao bound.
Advantageously, the switching design according to the present invention can be realized both for one-dimensional and for two-dimensional arrays including an arbitrary number of transmit antennas ≧2.
For the person skilled in the art, it will be understood that the described features of the present invention can be suitably modified and combined with one another without departing from the core idea of the present invention.
| Number | Date | Country | Kind |
|---|---|---|---|
| 10 2014 206 927.0 | Apr 2014 | DE | national |
