JP5410478B2 - Integrated packet detection in a wireless communication system with one or more receivers - Google Patents

Description

  Wireless networks are becoming increasingly popular because computers and other devices can be coupled for data communication without the need for wired connections between network nodes. One set of standards for wireless networks is the IEEE 802.11 standard, but other wireless standards or protocols may be used instead. There are at least two widely used standards in the IEEE 802.11 standard, 802.11a and 802.11b, and communication systems and devices need to support both standards and / or in the area where both are used. May need to work.

  Since both technical standards operate in different frequency ranges, interference between them can be avoided. However, recent additions, such as the 802.11g standard, allow for OFDM transmission in the 2.4 GHz band (802.11a is an OFDM transmission protocol) where there can be 802.11b direct spread spectrum transmission. Yes. Because it must detect the presence of a packet with a small preamble and indicate whether the packet is an 802.11a packet or an 802.11b packet with a very low error rate, packets with different modulations are in the same network The fact that it can exist presents difficulties in the design of 802.11 packet detection circuits.

  Before decoding bits with information content provided by the transmitter, the receiver typically senses the transmitted packet and then characterizes the channel, synchronizes with the transmitted packet, etc. Perform the steps. Packet detection determines that a packet is present on the channel (ie, the packet is transmitted by a transmitter or has already been transmitted) and performs further processing on the type of packet (at least with respect to the packet or its contents). To the extent required to perform) and activate the receiver components necessary to perform further processing. In some receivers, the receive logic is implemented by DSP instructions provided to a digital signal processor (DSP). If logic that performs data processing on received packets is implemented as instructions, those instructions may remain unexecuted until the packet detector indicates a received signal that contains a packet to be further processed. If the logic that performs the data processing of the received packet is implemented as a wired circuit configuration, the receiver will keep such circuit until the packet detector indicates a received signal containing the packet to be further processed. It may be configured to remove or reduce power to the configuration. In either case, processing power and / or computational effort is protected if no packet is detected. This saves power and / or processing requirements that are often imposed at the wireless receiver, except those required for packet detection. The receiver should detect the presence of the packet and take whatever action is required to initiate packet processing before the critical elements of the packet are lost, so that a portion of the packet is not lost. Hence, packet detection should be efficient and quick.

  In addition to the problem that 802.11 signals may overlap, 802.11 receivers may be narrowband non-802.11 signals such as Bluetooth, scientific equipment, medical equipment or microwave ovens. The receiver's packet detector should preferably not cause false triggers due to such effects. Aside from the interference and packet detection problems, the conventional 802.11a receiver operates at a 20 MHz sampling rate, while the conventional 802.11b receiver operates at a sampling rate of 22 MHz, so that 802.11a and A receiver that must receive and process both 802.11b signals cannot use a simple common sampling scheme.

  A typical node in a wireless network includes a receive chain and a transmit chain, each chain using only one antenna at a time. However, in a multiple-input multiple-output (MIMO) communication system, one or more transmission antennas and / or one or more reception antennas are used, and each of the transmission antennas sometimes has a different bit stream from the other transmission antennas. Each transmitting antenna is preferably receiving input from a channel that is at least slightly different from the other receiving antennas.

  MIMO communication systems are well known in the art. The system generally includes a transmitter with multiple (Mt) transmit antennas in communication with a receiver with multiple (Mr) receive antennas, and Mr and Mt may or may not be equal. In some modulation schemes, the transmitted data bits are grouped and each group of bits is mapped to a symbol (a specific phase and amplitude combination) in the constellation. Many arrangements are well known in the art, including two phase shift keying (BPSK), four phase shift keying (QPSK) and quadrature amplitude modulation (QAM) arrangements. In a MIMO communication system, each of the Mt transmit antennas transmits a symbol representing a different group of bits substantially simultaneously. Thus, if each symbol represents B bits, the number of bits transmitted per channel “period” is B * Mt.

  Each receive antenna receives a signal that is a combination of signals from the transmit antenna, transformed by channel properties (eg, fading and delay) and noise. The receiver decodes (i.e., reproduces) the Mt transmission signal from the Mr reception signal using knowledge about the symbols that can be transmitted and the properties of the communication channel. Because of the improved reception capability of multi-antenna systems, multi-antenna systems are expected to receive signals with lower signal-to-noise ratios (SNR) than other systems. As the expected range of operation for SNR becomes wider, correct packet detection with lower SNR is expected, which makes many conventional packet detection schemes inappropriate.

  It would be worthwhile to overcome the drawbacks of the prior art described above.

  In one embodiment for a wireless receiver, the packet detector combines 802.11a packets, 802.11b packets, and interference that is within the monitored frequency range but is not formatted as an 802.11a packet or an 802.11b packet. To detect. A packet detector can use signals from one or more antennas. The signal is detected using the differentially detected correlation. In addition to packet detection, the packet detector can ascertain the signal level, noise level, and location of narrowband interference. The processing of packet detection and other indication confirmation can be done simultaneously when a signal is received.

  A further understanding of the nature and advantages of the invention disclosed herein may be better understood by reference to the remaining portions of the specification and the attached drawings.

1 is a block diagram of a simple wireless network that can use the present invention. FIG. 2 is a block diagram illustrating a connection between one device and one network connection of the wireless network shown in FIG. FIG. 3 is a block diagram of a reception unit of node hardware that can be used in the hardware illustrated in FIG. 2. FIG. 4 is a block diagram of a portion of a receiver including packet detection elements similar to those that may be used in the receiver of FIG. FIG. 5 is a block diagram of an 802.11b packet detector that may be used with the receiver of FIG. Schematic illustration of correlation sampling. FIG. 5 is a block diagram of an 802.11a packet detector that may be used with the receiver of FIG. FIG. 5 is a block diagram of a continuous wave (CW) detector that may be used with the receiver of FIG. FIG. 5 is a block diagram of an interference detector (including 9A, 9B, 9C) that may be used with the receiver of FIG. FIG. 5 is a block diagram of an interference detector (including 9A, 9B, 9C) that may be used with the receiver of FIG. FIG. 5 is a block diagram of an interference detector (including 9A, 9B, 9C) that may be used with the receiver of FIG. FIG. 4 shows a series of plots showing the performance of one embodiment of the packet detector described above; showing an 802.11a detection curve. Fig. 3 shows a series of plots showing the performance of one embodiment of the packet detector described above; showing an 802.11b detection curve. FIG. 5 shows a series of plots showing the performance of one embodiment of the packet detector described above; showing a CW interference detection curve.

  FIG. 1 illustrates a simple wireless network that may use the present invention. As shown in FIG. 1, the wireless network 10 includes a plurality of nodes 12, and each of the nodes 12 can communicate with at least one other node 12 of the wireless network 10. In a specific implementation, the wireless network 10 is a local area wireless network as may be used in a building, campus, car or similar environment.

  In a specific embodiment, wireless network 10 is designed to be compliant with one or more IEEE 802.11 standards. However, it should be understood that other standard and non-standard networks can be replaced to solve problems similar to those solved in 802.11 environments. For example, the IEEE 802.11g standard expects a different signal than the 802.11a or 802.11b standard, and the 802.11 technical standards set may be further modified in later developments. Thus, although there are many examples described here that solve the problem of packet detection (and other tasks) in an environment where 802.11a and 802.11b packets exist and possibly other interfering signals, The teachings in the disclosure may be used in systems where two different protocol standards are used and with / without unwanted interference. In one example, at least one of the protocols is an extended 802.11a protocol suitable for use between devices that support the protocol.

  As shown, some nodes are coupled to the node device 14 while other nodes are coupled to the wired network interface 16. For example, node 12 (1) is connected to node device 14 (1), while node 12 (3) is connected to wired network interface 16. FIG. 1 represents a simplified and generalized wireless network diagram. The source of the interference signal is not shown but is assumed to exist.

  Examples of node device 14 are laptops, personal digital assistants (PDAs), or other portable or semi-portable electronic devices that need to communicate with other devices, or wired to a network or other device Includes fixed electronic devices that need to communicate with other devices where connections are not available or easily provided. The wired network interface 16 connects each node to the network. Examples of such networks include the Internet, a local area network (LAN), or a public or private connection to a TCP / IP packet network or other packet network or network.

  In normal operation, a plurality of node devices are provided with circuitry and / or software that functionally implements the node 12, and between such node devices and a network connected to a wired network interface. One or more network access points are provided in the wireless network 10 to provide access. In the terminology used herein, a node attached to a node device is referred to as a “station”, and a node attached to a wired network interface is referred to as an “access point”. One example of such a system use is to connect computers in a building to a network without having to draw network wiring to each computer. In that example, the building will be equipped with a fixed access point connected to the network that is within the wireless communication range of each wireless network card of the station connected to the network.

  FIG. 2 shows the connection of one device and one network connection in more detail. As shown in FIG. 2, the node device 14 is connected to the device I / O unit of the node hardware 20. The node hardware 20 includes a transmission unit and a reception unit, and each is connected to a device I / O unit. The transmitting unit transmits a signal to the receiving unit of the access point hardware 22 through the wireless channel 21. The receiving unit is connected to the network I / O unit, and thus provides a data communication path from the device 14 to the network 28. The path from the network 28 to the device 14 also includes the network I / O unit of the access point hardware 22, the transmission unit of the access point hardware 22, the reception unit of the node hardware 20, and the device I / O of the node 20. Provided by the department. The characteristics of the radio channel 21 are similar to the location of the node hardware 20 and access point hardware 22, for example, interference such as walls, buildings and natural obstacles, as well as other devices, transmitters, receivers. And depends on many factors such as the influence of the signal reflecting surface.

  In general, the node hardware 20 may be integrated into the device 14. For example, if the device 14 is a laptop computer, the node hardware 20 may be added to a PCMCIA card that is inserted into the PCMCIA slot of the laptop. In general, the access point hardware 22 is implemented as part of a wired network interface device that is used to simply connect a wired network to a wireless network. Regardless of the above general implementation, nothing is that FIG. 2 is completely symmetric, ie, node hardware 20 and access point device 22 are almost ideal cases of hardware devices. It should be understood that it does not interfere.

  The following is a detailed description of the receiving unit. FIG. 3 illustrates the components of the receiving unit 30. The receiving unit 30 receives signals on one or more radio channels via the antenna 32, and the signals are first processed by the RF unit 34. For example, the RF unit 34 may perform a process of forming a baseband signal on the signal in order to form a digital signal stream. As shown, the receiver 30 may include various subsections 40, 42, 44 for each of the FIR 35 and 802.11a, 802.11b, and enhanced 802.11 processes. The receiver 30 may include an integrated packet detector 37 described in detail below. Further details of the components of the receiver 30 that are not fully described in this specification can be found in US Patent No ._____ (filed on Feb. 5, 2002, "Multi-Antenna Wireless Receiver Chain With Vector Decoding" US Patent Application 10 / 068,360), which is incorporated herein by reference for multiple purposes. It should be understood that the present invention is not limited to the specific receiver implementation shown herein or in the above references.

  The integrated packet detector 37 performs the process of determining the start of the packet on the input signal and thus displays whether the packet detection signal is, for example, a subsection 40, 42, 44, in order to indicate whether any further processing is required. Can be output to other elements. If the receiver 30 is implemented as a digital signal processor instruction, the integrated packet detector 37 determines whether a packet is detected and whether the processor executes code for the other blocks shown. It may be a code for setting a flag used to determine FIG. 3 shows other signals derived from the same processing (an “auxiliary signal” used by other components similar to the integrated packet detector 37 and a “pass signal” generated by the packet detector 37 but not used directly. ) And possibly a plurality of detection signals supplied. The subsection 40, 42, or 44 may operate at a “unique” sample rate that is specific to the processing handled therein, but the integrated packet detector 37 is “unique” sample rate with respect to the protocol in which the packet is detected. If not, it can operate at a common sample rate such as 20 MHz. This allows for a more efficient integrated packet detector.

  FIG. 4 illustrates the integrated packet detector 37 in more detail. As shown in FIG. 4, I + Q inputs from one or more antennas are received at 802.11b detector 102, 802.11a detector 104, CW detector 106, power level detector 108, and interference detector 110. Input to a number of constructed detection blocks, each of which is described in detail below. The output of the detection block is provided to a controller 120 that provides an output for use at the receiver. In some embodiments, some functions described herein as being performed by the controller 120 may be performed instead on individual blocks. In some embodiments, additional detectors are provided for other subsections, such as extended 802.11 subsections.

In the embodiment shown in FIG. 4, the output of the block input to the controller 120 is as shown in Table 1. In the embodiment shown in FIG. 4, the output of the controller 120 is as shown in Table 2.

  A rough estimate for the frequency offset of the transmitter / receiver can be calculated elsewhere, but if it is calculated by the interference detector 110, the signal prior to the determination of the frequency offset that would reduce the detection sensitivity to narrowband interference. Any narrowband interference from can be easily removed. In general, narrowband interference is “removed” by removing subcarriers around the detected narrowband interference.

The cw_present and pwr signals may be used by the controller 120 as an indication of the presence of strong narrowband interference, in which case any 802.11a or 802.11b packet detection decision is ignored as being most likely false alarm .

  Although the controller 120 appears to be provided for the detection block in FIG. 4, the controller 120 may be provided as such, or may be responsible for other functions not shown or controlling the processing. Good. Some of the functions of controller 120 are described below with reference to blocks that generate inputs to controller 120. For example, the “detect_11a” signal is described in the following description of how the 802.11a evaluation index is generated. The pass signal is used in other blocks, such as 802.11a or 802.11b subsections, and need not be used by the controller 120 itself.

  In some implementations, the indicator signals (detect_11a, detect_11b, cw_present, etc.) may be binary signals, but in other implementations the probabilities, reliability, or accuracy of what the signal represents is represented. There may be multiple levels to display the value. For example, detect_11a may have a signal value indicating that an 802.11a packet is received with a probability of 70%. However, in the end, since the binary decision on whether to activate the 802.11a / b subsection needs to be made, the above indicator will eventually be expected to be attributed to a binary signal.

802.11b detection
The preamble of the 802.11b packet includes a synchronization bit that does not include data. The preamble part of the signal is used to detect the presence of a packet and to estimate the signal and channel parameters. FIG. 5 illustrates one implementation of the 802.11b packet detector 102. In this implementation, packet detection is done by correlating the inputs and checking the correlation results. The detector 102 prepares or detects a pair of evaluation indices, c11b (correlation evaluation index for 802.11b) and p11b (power evaluation index for 802.11b) for use in formulating detect_11b in the controller 12. The instrument 102 may provide a single evaluation index that is a function of c11b and p11b, such as, for example, p11b / c11b or other variations. As an example of the use of the evaluation index for determining detection, if Equation 1 is true for a certain threshold T_11b, it may be considered that a packet has been detected.

If the number of symbols during the generation of c11b differs from the number of symbols during the generation of p11b, those values may be taken into account in the calculation, as illustrated in Equation 2.

  The threshold inequality in Equation 2 normalizes c11b by NSYM-1 (NSYM is a fixed constant indicating the maximum number of symbols used in the calculation (the symbol is 20 samples and one Barker code)), and nsym Normalize p11b (nsym is less than NSYM and is the number of symbols received since the last reset of the detector). In one embodiment, T_11b = 0.4 and NSYM = 10. In the absence of noise, the left side of Equation 2 evaluates to 1 when nsym = NSYM if each of the 20 sample symbols is an exact copy of the previous group of 20 samples.

  Since c11b is based on differential correlation, the addend for calculating c11b is less than the number of symbols obtained. For example, if nsym = 3 (60 samples), c11b would be based on two differential correlations (the correlation between the first two symbols and the correlation between the last two symbols). The normalization factor may be nsym-1, but then false alarms are more common when nsym is much smaller than NSYM. Since the addend used to calculate p11b is the same as the number of symbols, not less than the number of symbols, p11b is normalized using nsym to obtain the average power per symbol.

  One possible process and method for generating metrics uses input signals from N receive antennas. The receiver input signal is represented as ri (k) where i = 0, 1, 2,..., N−1 in FIG. Unless otherwise indicated or apparent, the signal can be a complex signal, and operations such as addition and multiplication can be complex operations.

  The input signal passes through a Barker correlator 302, one per antenna. One advantage of the approach described here is that when the input signal is processed for packet detection, many processes can be replicated because the signals are sampled at a common sample rate. We have determined that 802.11 packet detection can be sufficiently accurate even at a sample rate of 20 MHz. Of course, once a packet is detected and found to be an 802.11b packet, the sample rate may be changed to 22 MHz to operate at a fixed sample rate of 802.11. Other variations of sample processing such as, for example, sampling at 22 MHz for both detections, sampling at 20 MHz or a multiple or fraction of 22 MHz, or rates that are not a multiple or fraction of 20 MHz or 22 MHz are useful in other situations. .

  In the illustrated implementation, the 802.11b signal is generated at 22 MHz, but the Barker correlator 302 operates at a sample rate of 20 MHz. To obtain Barker correlation, instead of 11 taps sampled at 11 MHz, there are 20 taps corresponding to the Barker code sampled at 20 MHz, so a 20 tap Barker code is a resample of an 11 tap Barker code. is there. For example, a resampled and quantized 20-tap sequence is expressed as "0, 3, -2, -2, 3, 1, 3, 0, -4, 1, 3, 0, 3, 0, 2,-. 3, 0, -4, 0, -4 ", but may be other appropriately resampled Barker sequences with some arbitrary delay shift. With a resampled Barker sequence, the entire packet detector can operate at a sample rate of 20 MHz, and the processing done for 802.11a detection is related to 802.11b detection and auxiliary and / or pass signal generation. It can be used similarly. The output of the Barker correlator is represented by Rbi (k). Note that other modified Barker sequences may be used instead of those described above.

  Each output of the Barker correlator 302 is processed to determine two differential correlations, DCb1 (n) and DCb2 (n). As shown in Equation 3-4, since the correlation output is multiplied by the delayed and conjugated correlation output, the output of the Barker correlator 302 is differentially correlated. The differential correlation is processed in correlation block 304.

  As shown, the correlation block 304 includes not only one adder and two accumulators 312 (one per differential correlation), but also a delay line 306, a conjugator 308, and a multiplier 309 (per antenna). One). The input to adder 310 is multiplied by the Barker correlator output from each antenna, which is itself delayed and conjugated. The output of adder 310, labeled “C (k)” in FIG. 5, is provided to accumulator 312, which in this example has twelve values for C (k). Accumulate.

  The number of values accumulated is at the cost of reduced SNR performance, but may exceed 12. A smaller number of values will result in a significant SNR reduction. One reason for such an effect is illustrated by FIG. 6 which shows an example of the real part of the correlation output. In practice, the signal is probably a complex signal with indefinite, unknown phase offset and frequency shift, but only the real part is shown to keep this example simple. The differential correlation has two outputs, one constituted by the correlation of the section A and one constituted by the correlation of the section B. Differential correlation removes any phase offset and converts the frequency offset into a fixed phase offset at the differential output, which is summed over multiple symbols to improve SNR. In this example, it can be seen that most of the signal power is concentrated in the A section.

  The overlap between the A and B intervals reduces the SNR loss that occurs when a pulse occurs just at the A and B interval boundary. If there is no overlap, the SNR loss will be 3 dB, so the 12 sample interval has adequate overlap. The larger the number, the greater the overlap, resulting in an SNR penalty because noise-only samples between signal pulses are included in the correlation. Selecting a section slightly wider than the ½ symbol duration gives a gain close to 3 dB for the use of all symbol sections.

  It is possible to use more sections than two sections. In the case of a narrow delay spread where most of the signal power is concentrated in a narrow interval, for example, each has a size slightly larger than ¼ symbol, causing duplication to improve the SNR. Four sections may be used.

  The particular 12 (or any number) values for each accumulator are determined by their respective valid inputs. As shown in FIG. 5, the accumulator 312 (1) is enabled by the start1_en signal, and the accumulator 312 (1) is enabled by the start2_en signal. As illustrated in Equation 3-4, start1_en enables accumulator 312 (1) for samples 1-12, and start2_en enables accumulator 312 (2) for samples 11-22 (sample 21). −22 is just the sample 1-2 of the next symbol).

The processing of the correlation block 304 is shown in Equation 3-4 below, where the superscript “*” refers to the conjugate, the multiplication is the complex multiplication, and n refers to the symbol number (integer).

  In this description, i refers to the antenna number, and for some N, i = 0, 1, 2,. In the example shown in FIG. 5, N = 3, but N can be greater than 1, 2 or 3. In this description, k refers to the sample number, and k = 1, 2,. Since there are 20 samples per symbol for 802.11b detection (+2 due to overlap), S = 22 in this example.

  The differential detection present in Equations 3 and 4 (multiplying each sample by a conjugated sample at the corresponding position in the past symbol) tends to remove the effect of any frequency offset. Note that the differential correlations DCb1 (n) and DCb2 (n) are carried over to the overlapping portion of the symbol (eg, the 11th and 12th samples appear in the sum of both). This promises that there will always be one differential correlation with maximum signal power and minimum ISI regardless of data transmission over 802.11b signals.

  In 802.11b signals, there is a sign inversion caused by data modulation in the 802.11b preamble. To eliminate this effect, at block 314, the differential correlation is multiplied by its real sign and accumulated with several symbols. The accumulation is a moving sum, and the moving sum S1 for DCb1 (n) is the sum of the most recent NSYM-1 symbol values for a certain integer NSYM. If there are fewer than NSYM-1 symbols available since the last reset of the detector, a preferred value for NSYM = 10 and a value less than NSYM-1 may be used. The same applies to the moving sum S2 corresponding to DCb2 (n).

  Once the two moving sums are available, the comparison block 320 determines which moving sum has the larger absolute value, and the absolute value of the moving sum and the first or second moving sum. Outputs an index indicating whether was selected. The selected output sum is the evaluation index c11b described above. As described above, this evaluation index is used together with the power evaluation index to determine whether an 802.11b packet is detected. The power evaluation index p11b is calculated by summing the squares of the absolute values of the outputs of the Barker correlators 302 for all N antennas (or all active antennas) and collecting the moving sums for NSYM symbols. Is done.

  Using the above-described method of 802.11b packet detection or other methods, once an 802.11b packet is detected, the phase of the selected (maximum) moving sum value is the estimated offset with respect to frequency, Used as df_11b. If an 802.11b packet is detected, another value, peak_11b indicating a sample number that becomes a signal peak, is also calculated. In particular, the peak block 340 operates based on an index indicating the selection of one of two moving sums and the total output of differential detection.

  Peak block 340 sums the samples for NSYM-1 symbols from adder 310 in the moving sum unit. The moving sums from comparison block 320 are then added (subtracted if the selected moving sum is inverted in processing block 314 to account for sign reversal) and then accumulated over the 20 samples. From there, the correlator output in which the 8-sample cyclic moving average for the most recent NSYM-1 is detected differentially is extracted. For example, the correlator output for each of the 20 samples is stored in the sample buffer, and the output is accumulated for each of the symbols such that each of the values in the sample buffer is an accumulation for a different sample location. Assume. Thus, the first value in the sample buffer is an accumulation of correlations up to NSYM-1 for the first sample of the symbol (or the sample assigned to be the “first” sample). The content of the sample buffer is, for example, {8, 4, 1, 1, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 6, 2, 2} If so, the cyclic moving average for sample k is the average of the samples starting from the kth input. In the above example, the average of the 8 values starting from sample 18 is greater than that of any other consecutive 8 values (the 8 values starting from sample 18 are the same for samples 18-20 and 1-5) A), k-18 yields the maximum value. The maximum value of the sample is used as the peak, and the peak block 340 outputs an index for that value (eg, the value “18” for the above example) for use as an 802.11b symbol timing reference or for other purposes. To do.

802.11a detection
FIG. 7 is a block diagram of the 802.11a detector 104. The basis of packet detection is a correlator that correlates the input signal with 16 samples of 802.11a short training symbols. Through the correlator, differential detection is performed to eliminate the effect of frequency offset between the transmitter clock and the receiver clock. The differentially detected correlation output is then accumulated for a plurality of symbols. Correlation may be done using a fixed inverse-correlation pattern for the signal for one or more antennas, rather than differentially correlating with the received signal, but for low SNR signals, differential correlation Usually provide better results.

  In one possible process and apparatus for 802.11a metrics generation, input signals from N receive antennas pass through an OFDM correlator 402. In FIG. 7, for the k th sample of the i th antenna, the input signal is represented as ri (k) and the output of the corresponding OFDM correlator 402 is represented as Rai (k). To reach the power level metric p11a, the absolute value of the square of Rai (k) is summed by adder 404 and then summed for the most recent NSYM samples.

Each output of the OFDM correlator 402 is processed to determine two differential correlations by multiplying each OFDM correlator output with its own delayed one to make a differential detection, and another addition. The summation unit 410 sums for all OFDM correlators (ie, for all antennas). Using the correlation block 405 as shown in FIG. 4 results in the processing shown in Equations 5 and 6 below (superscript “*” refers to conjugation, multiplication is complex multiplication, and n is Referring to the symbol number (integer), S = 18 for 802.11a detection (16 samples per symbol, +2 sample overlap).

  The differential correlation is carried over to the overlapping portion of symbols. This promises that there will always be one differential correlation with maximum signal power. Once these differential correlations, DCa1 (n) and DCa2 (n), are calculated, the moving sum of these values is stored in NSYM-1 symbols (detector recently reset using accumulator 414). If so, you can get less). The moving sums are compared (in comparison block 416) to determine the one with the largest absolute value, and the larger absolute value is output as the c11a evaluation indicator signal. If fewer than NSYM-1 symbols are available since the last reset of detector 104, the moving sum is for the received (symbol) samples.

In one embodiment, the controller 120 indicates whether an 802.11a packet is received through the signal detect_11a. One binary display is whether or not Expression 7 is satisfied with respect to a certain threshold value T_11a and a certain threshold value T_11ab. In one embodiment, NSYM = 6, T_11a = 0.4, and T_11ab = 0.5.

In an embodiment in which detect_11a is a multilevel signal, when c11a / c11bb> T11ab, detect_11a may be an amount commensurate with the amount shown in Equation 8.

CW detection
802.11a and 802.11b detectors perform well for additional white Gaussian noise, but generally distinguish between desired packets and narrowband interference such as Bluetooth and other interference types Can not. Incorrect detection may cause processing to start others that do not involve listening to the start of the packet, and between the receiver's incorrect start processing and failing the incorrect start and resetting False triggers are undesirable because true packet start may be missed. In order to prevent false triggering based on narrowband interference, a narrowband interference detection circuit is used. An example of such a component is shown in FIG.

  CW detection is preferably performed each time one of the packet detectors 102 or 104 indicates that a packet has been detected. The CW detection may be performed simultaneously with other processes, or may be performed on a set of buffered samples after packet detection occurs.

  In order to minimize hardware complexity, narrowband interference detection may be done on one active antenna. Since the main purpose is to detect strong interference in the absence of the desired signal, no significant improvement in the performance of this step can be expected with more than one antenna. However, it is possible to use more than one antenna.

  As illustrated by the components shown in FIG. 8, the input signal is buffered to make the last 24 samples available to the packet detector available. A window filter 604 is applied to the 24 samples, and a raised cosine window with a roll-off length of 8 samples is applied to the first 8 and last 8 samples. This tends to minimize “leakage” of narrowband interference power to adjacent FFT outputs, thus improving detection probability. However, other window schemes may be used instead. The window filter 604 folds the over sample by adding samples 1-4 to samples 17-20 and samples 21-24 to samples 5-8, resulting in a set of 16 samples. In one implementation, the effect of the window filter 604 is predetermined, and a set of 16 functions for the inputs is defined and applied to the 24 inputs, resulting in 16 values for the output of the window filter 604.

  On the other hand, when the window filter output is obtained, the 16-sample FFT 606 performs FFT on the window filter output and supplies it to the power detector 608 and the maximum value detector 610. The maximum value detector 610 determines the FFT output sample (mth sample) having the maximum power.

  Further processing is performed by blocks 612 and 614. If the combination of the m th sample and the two samples on each side of the m th sample is greater than the threshold times the power of all samples (or “other samples”), narrowband interference is expected. And the CW_present signal is asserted. When the controller 120 receives the CW_present signal, the 802.11a or 802.11b trigger is ignored. The CW_present signal may be binary, but it may be multi-level indicating the probability, reliability, or accuracy value associated with the determination.

  The above technique can be used with one antenna, but can be used with multiple antennas. If multiple antennas are enabled and used, there may be multiple instances for FFT 606 and power detector 608, the outputs of multiple FFTs being fed to multiple power detectors, and the power detector outputs added together. Adapted.

power
In some implementations, not all antennas are used to ensure power when securing power is more important than obtaining better performance using all antennas. The antenna used is referred to herein as an “active” antenna. The power component 108 estimates the received signal power for each antenna, and the estimated received signal power is used for automatic gain setting (AGC) of the RF receiver. Changes to RF gain occur in each of the following three cases:
1) If one I-sample or one Q-sample for one of the active antennas is taken: in this case, the RF gain of all active antennas can be reduced by a coarse gain step (eg 20 dB) .

  2) If the total power of all active antennas is below a certain threshold: In this case, the RF gain of all active antennas is a coarse gain step (generally the rising and falling steps are the same, (Which may be different).

  3) When a packet is detected: In this case, a power adjustment is made based on the estimated power averaged over the last 32 samples.

  The power of all active antennas may be adjusted in 4 dB steps so that the resulting backoff is approximately equal to some target backoff, and the gain may be changed after every packet detection. Inactive antennas are set to a higher backoff level to prevent clipping that may occur if these antennas happen to have significantly more power than active antennas. When more than one active antenna is used in AGC, programmable parameters limit the maximum gain difference between different antennas. The above limitation is made to prevent this, since the gain of an antenna receiving only noise may cause problems with 802.11b receiver training or other situations.

  Although FIG. 4 shows only “pwr” output from the power component 108, the power component 108 may output a gain setting for each antenna. For example, once a packet is detected, the controller 120 may send a gain setting (and possibly a pwr setting) to the RF receiver to update the RF receiver gain setting.

Interference position
Interference detector 110 provides additional information to assist the receiver. The information includes the frequency position of the expected narrowband interference, the coarse frequency offset (df — 11a) between the transmitter and receiver for the 802.11a signal, signal power and noise power. FIG. 9 shows one possible circuit (including components 110A, 110B, and 110C) for such position and other indicator determination.

  As shown, 32 samples from each antenna are processed to determine the index. Samples are obtained from the sample buffer 602. Although only one circuit is shown in FIG. 9A, it should be understood that the circuit for the i th antenna is shown and multiple instances may exist for multiple antenna receivers. It is. Alternatively, the circuit may be used continuously for more than one antenna (eg, if the circuit is implemented by instructions for a digital signal processor) if there is no performance problem, The circuit may be provided in one antenna, and all the circuits may be implemented by dedicated hardware such as a custom ASIC.

  From the 32 samples in the sample buffer 602 to form the frequency signals Fi (k) and Fia (k) (k represents one of 16 subcarriers k = 0,..., 15) The FFT block 604 performs FFT on the first 16 samples, and the FFT block 606 performs FFT on the next 16 samples. The output of the FFT block 606 is conjugated by a conjugator 608, and the outputs of the FFT block 604 and the conjugator 608 are multiplied by a complex multiplier 610 to form a multiplier output Zi (k). An accumulator 612 sums the multiplier outputs for 12 samples (k = 1..6, 10..15) to form a correlation output Ci. The particular sample used in the above example is based on the short training subcarrier present in the first 8 microseconds of the 802.11a preamble. Other (or fewer than all the samples used in the above example) may be used for the purpose of modifying this process. The subcarrier of k = 0 is not used because it is a DC component.

  In the example shown in FIG. 9, since the number of antennas is N = 3, the component 110A may be illustrated three times, and F1b (k), F1a (k), Z1 (k), C1, F2b (k), Generate values for F2a (k), Z2 (k), C2, F3b (k), F3a (k), Z3 (k) and C3. Those values are used in components 110B and 110C as shown in FIGS. 9B and 9C, respectively.

  As shown in FIG. 9B, the correlation output values for all antennas to be processed are accumulated by accumulator 620 and the angle of the complex value output from accumulator 620 is calculated by angle block 622. The multiplier output values for all antennas processed are summed by another adder 621. The output of angle block 622 and adder 621 is output by rotation block 624 to determine a phase corrected value Z ′ (k) equal to the input sample Z (k) rotated by the angle detected by block 622. Used.

  The maximum value detection block 626 receives Z ′ (k) and outputs the maximum Q value as a subcarrier representing the interference position output. Blocks 628 and 630 define a coarse frequency offset df_11a from the sum of multiplier output values excluding subcarriers corresponding to the interference location and its neighbors (one for each side).

  The component 110C shown in FIG. 9C relates to one antenna (or multiple antennas if so usable), the i-th antenna. As shown in FIG. 9C, the frequency signals Fib (k) and Fia (k) are provided to an absolute value block 640 that outputs the power (or at least an expression that matches the power) of the frequency signal. The signals are added by an adder 642. An accumulator 644 sums the results for 16 samples. The accumulator result is rounded to a power of 2 by rounder 645, and the output of rounder 645 is the signal scaling factor for the i th antenna. The accumulator result is also used for noise scaling, and adder 646 subtracts from the accumulator result a double of the correlation output value Ci for the i th antenna. Another rounder 648 rounds the result from adder 646 to a power of two, which is defined as the noise scaling factor for the i th antenna. Signal strength and noise scaling factors are used for multi-antenna combining.

  If interference and a valid packet are detected, the interference is removed and the packet may be processed to obtain data, but if there is a detected interference and no valid packet is detected, the other There is no need to do anything. For example, narrowband filtering using the interference position, such as signal filtering using a band-gap filter at the interference position, may be performed. Another approach is to pass the value to a Viterbi decoder that displays a low confidence in the value evaluated at the frequency to which the interference belongs. Interference detection may be utilized in other components of the receiver other than the packet detector. For example, the interference position can be used in a synchronization process for ignoring spurious signals.

Experimental result
FIG. 10 illustrates certain experimental results using the above circuit for index detection and creation. Each of these plots is for one active antenna and plots the probability versus threshold in the presence of additional white Gaussian noise.

  FIG. 10A illustrates 802.11a packet detection results including false alarms (packet not present but packet detected) and detection failure (packet present but no packet detected). The curve labeled “FA” represents the probability that a packet is erroneously detected within a single symbol interval. The other curve is a detection failure curve for the signal-to-noise ratio indicated by the labels (-3 dB, 0 dB, 3 dB and 6 dB). In practice, detection decisions are made for each symbol duration, so the probability of false alarms is higher than the probability shown in FIG. 10 within a certain number of symbols. Since false alarms may lead to detection failure because the receiver will be “listen” for a certain period of time, the selection of a detection threshold with a sufficiently small false alarm probability is important. For 802.11a, a reasonable threshold is -4 dB.

  FIG. 10B illustrates 802.11b packet detection results including false alarms and detection failures using the same notation and SNR values as in FIG. 10A. Similar to 802.11a detection, within a certain number of symbols, the probability of false alarm is greater than the probability shown in the figure, and the detection threshold should be chosen with a sufficiently low false alarm probability. A reasonable threshold for 802.11b is -5 dB.

  FIG. 10C illustrates a CW interference detection curve. The solid line labeled “FA” represents the false alarm probability, and CW interference is detected on the OFDM preamble with a delay speed of 250 ns. A false alarm means that a valid OFDM preamble is rejected even though it can be received without interference. Dashed lines represent detection failure probabilities for signal-to-interference ratios of 0 dB, −10 dB, −20 dB, and −30 dB as displayed.

  The above description is illustrative and not restrictive. Many variations of the invention will become apparent to those skilled in the art upon review of this disclosure. The scope of the invention should, therefore, be determined not with reference to the above description, but instead should be determined with reference to the appended claims along with their full scope of equivalents.

Claims (18)

In a packet detector for detecting data packets of a first format and a second format,
Comprising inputs from a plurality of antennas configured to receive different signals;
The input is coupled to an analysis circuit, a first subsection and a second subsection;
The analysis circuit has a first ratio of a maximum first differential correlation moving sum of the different signals to a moving sum of the first power levels of the different signals for the different signals received from the input. In response to satisfying a first predetermined condition, it is determined that there is a data packet in the first format, and the maximum first of the different signals relative to a moving sum of the second power levels of the different signals is determined. second ratio of the moving sum of the differential correlation in response to satisfy the second predetermined condition, be configured for determining a data packet is present in said second format,
The first format corresponds to a first protocol standard;
The second format corresponds to a second protocol standard different from the first protocol standard;
Packet detector. The analysis circuit includes:
A correlation circuit configured to determine a correlation evaluation criterion for the first format;
The packet detector of claim 1, comprising power measuring means configured to display a first power level of the different signals received through the plurality of antennas. Based on the fact that the first ratio of the moving sum of said maximum of the first differential correlations for running sum of the first power level satisfies the first predetermined condition, the first The packet detector of claim 1, further comprising a detector output configured to output a signal indicating the presence of a data packet in the format.   The packet detection of claim 3, wherein the first subsection is configured to receive the signal indicating the presence of a data packet in the first format and processes the data packet based on the reception of the signal. vessel.   The packet detection of claim 3, wherein the second subsection is configured to receive the signal indicating the presence of a data packet in the second format and processes the data packet based on the reception of the signal. vessel.   The packet detector of claim 1, wherein the analysis circuit operates at a single sampling frequency.   The packet detector of claim 6, wherein the single sampling frequency is about 20 MHz.   The packet detector of claim 6, wherein the single sampling frequency is about 22 MHz. The analysis circuit of claim 1 , wherein the analysis circuit is configured to determine that a data packet is present in the first format in response to the first ratio exceeding a first predetermined threshold. Packet detector. In a method for detecting data packets in a first format and a second format,
Receiving multiple signals through multiple antennas;
A first subsection for processing data packets encoded in the first format; a second subsection for processing data packets encoded in the second format; and an analysis circuit And supplying the plurality of signals;
Depending on the first ratio of the maximum of the first running sum of the differential correlation of the plurality of signals for moving sum of the first power level of the plurality of signal satisfies the first predetermined condition Te, wherein the first format is determined that the data packet exists, a second moving sum of the maximum of the second differential correlation of said plurality of signals for running sum of the second power level of the plurality of signals Analyzing the plurality of signals including determining that a data packet is present in the second format in response to a ratio satisfying a second predetermined condition ; and
The first format corresponds to a first protocol standard;
The second format corresponds to a second protocol standard different from the first protocol standard;
Method.   The method of claim 10, wherein the analysis circuit operates at a single sampling frequency of about 20 MHz. Based on determining that the first ratio of the moving sum of said maximum of the first differential correlations for running sum of the first power level is greater than a first predetermined ratio, said first 11. The method of claim 10, further comprising transmitting a signal indicating the presence of a data packet in one format to the first subsection.   The second subsection is configured to process the data packet based on the signal indicating that the data packet is encoded in the second format. 12 methods.   13. The method of claim 12, wherein processing by the first subsection is performed in response to a signal from a detector output indicating that a data packet is present in the first format. Analyzing the received signal
Determining a correlation evaluation criterion for the first format;
Measuring the first power level of the plurality of signals.   The method of claim 10, wherein the analysis circuit operates at a common sampling frequency.   The method of claim 10, wherein the analysis circuit operates at a common sampling frequency of about 22 MHz. Said first predetermined condition, the first ratio is that exceeds a first predetermined threshold value, The method of claim 10.

Images