阅读说明：本技术 低功率广域网中的相对跳频 (Relative frequency hopping in low power domain networks ) 是由 O.B.A.塞勒 N.索尔南 于 2019-07-01 设计创作，主要内容包括：低功率广域网中的相对跳频。用于IoT应用的跳动扩展频谱无线网络,所述网络包括移动设备,所述移动设备具有未经同步的本地频率基准。所述传送器以这样的方式使用在相对不同的频率方面限定的跳动序列：使得接收器可以确定传送的跳动序列,而无论大频率误差的存在。(The transmitter uses hopping sequences defined in terms of relatively different frequencies in such a way that a receiver can determine the transmitted hopping sequence despite the presence of large frequency errors.)

A wireless communication network of the kind , comprising a plurality of radio transmission devices in a frequency band, the frequency band being divided into a plurality of channels, wherein,

the transmitting devices each have a local frequency reference and are arranged to: modulating a carrier and processing the modulated signal into a spread spectrum radio signal by switching the carrier frequency between a number of hopping frequencies in said frequency band according to a hopping sequence, and,

the conveyor is arranged to: determining a hopping sequence by repeatedly incrementing an initially hopped channel according to the determined incremental succession, thereby obtaining a sequence of channels, each channel in the sequence defining an element of the hopping sequence; and transmits radio signals over the radio interface.

2. The wireless communication network of claim 1, wherein the channels do not overlap.

3. The wireless communication network of claim 2, wherein in the determination of the hopping sequence, the transmitter applies a modulo operation after each increment to maintain the frequency hopping within a predetermined limit.

4. The wireless communication network according to claim 1, comprising at least receiving gateways arranged for detecting a signal transmitted from a transmitting device at an initial receiving frequency, forecasting frequency hopping by repeatedly incrementing the initially received frequency according to a determined incrementing succession, receiving a signal at the forecasted frequency hopping.

5. A wireless communications network according to claim 4 when dependent on claim 2, wherein the receiving gateway is arranged to determine when a hop frequency expected to fall within the vicinity of a predetermined limit can be at any of two different values, depending on whether the increment results in a frequency which is judged by the transmitter to be outside the predetermined limit, and to listen at two different frequency values.

6. A wireless communications network according to claim 3 when dependent on claim 2, wherein the transmitter device is arranged to: omitting transmissions whose hopping frequencies fall into hops closer to a predetermined limit than a determined threshold.

7. The wireless network of claim 1, wherein the frequency band is divided into sub-bands; the generation of the sequence of hops in the transmitting device includes a determination of a sequence of subbands and each hop is transmitted into a subband defined by the sequence of subbands at a channel defined by the sequence of channels.

8. A wireless network according to claim 1, wherein the transmitting device is arranged to select a hopping sequence among a plurality of possible sequences.

9. The wireless network of claim 8, wherein the selection of the hopping sequence is based on a synchronization status of a frequency reference of the transmitting device.

10. The wireless network of claim 8, wherein the selected hopping sequence is signaled by a selection of a frequency used at .

11. A wireless network according to claim 1, wherein the transmitting device is arranged to signal the values of hopping frequencies as synthesised on the transmitting device's own frequency reference.

12. The wireless network of claim 11, wherein the value is implicitly signaled by a selection of the th sub-band.

13. The wireless network of claim 11, wherein the value is explicitly signaled by including modulated data in a preamble that specifies a channel or frequency used for the th hop or any hop in the sequence.

Technical Field

Certain use cases of the present invention relate to low power measurement nodes for IoT (internet of things) applications, and to IoT domain networks that include multiple receiving gateways in addition to measurement nodes, although the present invention is not limited to those applications.

Background

Local area networks, such as WiFi and bluetooth, have been used successfully in applications, but they require a local infrastructure that is not always available or desirable to be connected to the internet, and are hardly applicable to mobile applications where the sensor nodes can move outside the accessible range of WiFi or bluetooth gateways.

1. If, as is usually the case, the network is operated in an unlicensed band, a high level of interference resistance is necessary.

2. Compliance with the protocol must be ensured.

The LoRa network employs chirped spread spectrum modulation and has, among its strong points: low hardware complexity, no need for accurate frequency reference or timing, easy synchronization and positioning. However, there are certain limitations in capacity for low data rates.

Sigfox technology and ultra-narrowband technology suffer from very high collision rates in comparison to system load in ultra-narrowband networks, collisions occur in three dimensions: time, frequency, and power.

In summary, if the metric of weakest node performance is monitored, these effects limit the system load to 1% or less. This situation cannot be remedied by a simple emergency means of adding more receiving gateways.

The present invention proposes novel transmitter devices and corresponding wireless networks that implement a modified form of ultra-narrow band modulation, allowing the above limitations to be overcome or alleviated.

Disclosure of Invention

According to the invention, these objects are achieved by means of the objects of the appended claims.

Drawings

The invention will be better understood by means of the description of an embodiment given by way of example and illustrated by the accompanying drawings, in which:

fig. 1 is a simplified representation of an -domain low-power network including sensor nodes and gateways according to the present invention.

Fig. 2 illustrates a data frame used by the transmitter and receiver of the present invention, the data frame including a plurality of frequency hops.

Fig. 3 shows the subdivision of the useful radio spectrum in sub-bands at the transmitter and receiver end.

Fig. 4 shows a possible subdivision of subbands in a channel.

Fig. 5 shows a frequency plan with contiguous subbands.

Figures 6 and 7 plot the miss detection rate in an ALOHA network-whether or not it includes features of the present invention-and the packet error rate as a function of network load.

Fig. 8 shows a spread spectrum transmission preamble and data section, each split into multiple hops.

Fig. 9 illustrates an embodiment, where hopping frequencies are shifted by fractions of a nominal step.

Fig. 10-13 illustrate repeated collisions in the preamble under different hypotheses.

Fig. 14 shows a mechanism for compensating for the frequency error of a transmitter in hopping spread spectrum transmission.

Figure 15 compares frequency hopping transmissions by nodes with a low quality frequency reference with the same transmissions by nodes with a stable frequency reference.

Fig. 16 relates to an embodiment, where the kind of beat sequence is signaled by the selection of beats.

Detailed Description

Fig. 1 shows in a simplified manner a low power wireless network comprising a number of sensor nodes S0, S1, S2 and receiving gateways G0, G1. In a typical IoT application, the sensor nodes S0, S1, S2 would be simple battery-powered devices that acquire or compute data and upload (e.g., arrow 80) them to the gateways G0, G1. Although not required, downlink traffic (e.g., arrow 50) from the gateway to the node is also possible.

The number of sensor nodes is not limited and may exceed thousands in a practical use case the number of gateways is also not constrained and is indicated by the need to cover the area in which the sensor nodes are found.

The uplink communications 80 between the sensor nodes and the gateway use narrow or ultra-narrow bandwidth modulation and frequency hopping spread spectrum, the modulation is preferably forms of coherent phase modulation with a constant envelope, such as GMSK, MSK or PSK.

The uplink communication 80 from the sensor node to the gateway comprises a synthesis of a modulated radio signal, which is based on a local frequency reference in the sensor node. Due to cost and power considerations, sensor nodes cannot be equipped with high quality oscillators. Thus, the frequency of the uplink radio signal is affected by considerable errors, which may exceed the narrow bandwidth of the signal.

In contrast, the receiving gateway has considerably more computational resources, continuous power and an accurate time reference, such as for example a GPS-regulated clock, than the sensor nodes. They are preferably interconnected and can cooperate between them.

To conserve network capacity, the downlink communication may be cast to be received by all sensor nodes within range of the gateway, but point-to-point transmission is also possible. functions of the downlink message are synchronization of the sensor's time reference to the gateway time reference, which may advantageously be accomplished by, for example, LoRa cast packets.

In most IoT applications, the sensor nodes have only very limited power and computing resources. Random access to the radio channel, which means that the sensor node sends a message whenever it needs to send it, without signaling its intention or whether the listening channel is free, is therefore advantageous. These protocols, also known as ALOHA protocols, are sensitive to inter-system collisions: the capacity of the ALOHA protocol is limited by such collisions.

The transmitter of the sensor node switches the carrier frequency between several hopping frequencies in the available radio band according to an sequence so that the uplink transmission 80 comprises a series of successive hops with different carrier frequencies the frequency changes at each hop boundary preferably the hop length is considerably smaller compared to the message length- messages or frames comprise or several hops.

The system may have only predetermined hopping sequences, and in this case, messages with different start times will not collide in frequency, to the extent that their respective hopping sequences are time-shifted, or have multiple hopping sequences between which the transmitter can select.

Fig. 2 illustrates an uplink data frame or message, such as may be used in embodiments of the invention, the frame starts with a preamble containing a synchronization signal for detection at the receiver end and a physical header that includes information in the receiver for predicting and following a hopping sequence, the preamble preferably containing an identifier of a sequence that has been selected for subsequent data in the same frame or message if the system foresees more than possible hopping sequences.

The preamble may also include other information about the data format, such as an indication of the data rate, the modulation scheme used, and so on.

In order to prevent preamble loss, the information is preferably repeated several times in successive hopping fig. 2 shows a preamble containing three hops, but this is not a necessary feature of the invention.

The preamble is followed by a header that informs the receiving gateway about the nature and format of the subsequent payload. Importantly, the header and payload are interleaved and encoded for error correction. In the example, the header and payload include eight, respectively 32 hops, but these numbers may vary depending on the use case. Each beat contains a given number of modulation symbols and a corresponding number of modulated message bits.

To cite the figures, assume that the sensor nodes S0-S2 are equipped with low-level crystal oscillators with an error of 30 ppm, and modulate the data with a bandwidth of 400 Hz in the sub-GHz ISM band, one can expect a maximum error of 30 kHz in a defined carrier frequency, which is 80 times the modulation bandwidth.

Two possible ways of the foregoing are:

1. before transmitting, the reference of the sensor is synchronized with the gateway reference in the downlink.

2. The hopping sequence and the available frequencies are arranged such that frequency synchronization errors can be corrected and accommodated.

Sub-band and channelization

The hopping may occur within a sub-band or across sub-bands, any given frame is transmitted in a given channel in a sub-band, and since the frequencies synthesized in the sensing node are not directly related to those measured by the precision reference, it is convenient to define their frequencies by sub-band indices and channel indices inside the sub-band specified by the sub-band indices.

The use of several sub-bands is advantageous in many respects.

1. A higher extension in the frequency domain gives better coexistence in the unlicensed band, both as victim and aggressor. Regulatory constraints are simpler to satisfy and may allow for more frequencies and/or longer transmission times.

2. Higher diversity versus propagation/fast fading

3. Implicit signaling of the location from the th hop, th hop in the hopping sequence can be selected between a number of possible alternative starting frequencies and by this selection the transmitter transmits an information element.

4. Network planning and/or adaptive data rates.

It is useful to define these bands differently at the receiver and at the transmitter due to frequency mismatch between the receiver and the transmitter. We introduce three concepts: nominal, or regulatory, subband definitions, receiver subband definitions, and transmitter subband definitions.

If the transmitter has the possibility to select between several possible hopping sequences, the selected can be implicitly signaled by the selection of the frequency used at .

The selection of the hopping sequence can be based on the synchronization status of the frequency reference of the transmitting device.

Fig. 3 illustrates three subband definitions. In this example, the available radio spectrum includes three disjoint sub-bands: b0, B1 and B3, which are of equal width and regularly spaced in this example, but this is not an essential feature of the invention: the number, width and spacing of the sub-bands are arbitrary and they may be touching or overlapping depending on the implementation; they may correspond to different regulatory regions of the electromagnetic spectrum, but this is not essential. The nominal subband (NOM row) is the boundary of the frequency, measured by an accurate instrument, within which the transmission should be contained. The Transmit (TX) sub-bands are expressed in terms of frequency as determined by the frequency reference of the transmitter, i.e. they shift when the local oscillator shifts. To comply with nominal subband boundaries, the sensor node adjusts the transmitter subbands, reducing them so that the radio signal always falls within nominal limits even if frequency mismatch is considered.

Several strategies can be devised, but the general principle of is that the higher the frequency error indicated by the synchronization state, the narrower the subband will be adjusted.

The synchronization state may be obtained from a nominal error of the local frequency reference, from a drift model that may also include crystal temperature, from results of synchronization following the downlink, from the time elapsed since the last downlinks that follow in synchronization, or from a combination of all or of these elements.

In a case where the receiver is poorly synchronized (e.g., in a narrowband downlink transmission, or when GPS synchronization is at a failure), the receiver may tune to a wider set of frequencies, as shown, so it can receive virtually all transmissions within the nominal sub-band.

The nominal sub-bands as defined above may be divided into a suitable number of channels (the drawing shows only a reduced number.) the same sub-bands as defined for the transmitter will be slightly narrower due to the above adjustments and divided into the same number of channels however, due to frequency mismatch, the channels defined for the transmitter do not exactly correspond to the nominal channels.

In a sequence of hops, each hop is transmitted on a different carrier frequency that can be specified by a subband index and by a channel index within a subband. Preferably, the subbands are designed such that the correspondence between transmitter-nominal-and receiver is unambiguous, i.e. the presence of a carrier for the Transmitter (TX) in a given subband will be the same subband for the receiver, regardless of the synchronization state, since the separation between subbands is larger than the maximum expected frequency error. However, this is not the case for the channel, and the correspondence between the channel index selected by the transmitter and the actual frequency perceived by the receiver is not direct.

The frequency band may be split into sub-bands, each comprising a plurality of channels, otherwise all channels may be included in common divisions.

The transmitter of the invention may be arranged to determine frequency hopping by: the initially hopped channel is repeatedly incremented according to the determined succession of increments to obtain a sequence of channels, each channel in the sequence defining a hop. Preferably, the incrementing is followed by the modulo operation to maintain the frequency hopping within predetermined limits.

To the extent that the frequency of a given hop (or equivalent channel) can be derived from the previous frequencies by shifting, or shifting, followed by a modulo operation, the sequence only defines the relative frequency shift, or interval, associated with each hop, without constraining the channel from which the sequence starts.

In reception, , once the receiver detects a signal at the initial frequency (or channel), it can forecast the frequency hopping by repeatedly incrementing the initial frequency according to a known incremental succession, and successively tune in the frequency hopping to receive the entire message.

The offset does not depend on the frequency used but may depend on the beat index, e.g. it may be linearly increasing, or it may be generated from a pseudo-random sequence or permutation, which is known to the receiver or is algorithmically reproducible.

Implicit and explicit signaling of hopping sequences

The communication network of the present invention may use common hopping sequences for all uplink transmissions, or preferably multiple hopping sequences between which the transmitting node can select.

The transmitter side synthesizes a radio signal having a frequency determined exclusively with respect to its own time reference, but the frequency received by the receiver is undetermined since the reference has an undetermined error.

The receiving gateway can determine the th hop sub-band index unambiguously within the extent that the sub-bands are sufficiently separated, and also know the hop sequence index because this information is explicitly signaled in the header, but it cannot determine the channel index selected by the transmitter for the th hop unambiguously, and therefore cannot determine the step evolution of the hop sequence exactly.

To address this problem, the transmitter may have a stable frequency reference, such as a TCXO, and synchronize its frequency from the downlink frame or beacon in this case accuracy may be better than half or even quarter of the modulation bandwidth.

In another possible embodiment, the transmitter has a less stable frequency, such as a low-level XO, but its error is at least characterized and included within known limits.

To allow the receiver to determine the sequence of frequency hops, the transmitter sends out additional information on the channel used for the th hop or for the determined hops in the sequence, e.g., the th data hop following the preamble, despite the frequency error.

In a simple explicit signaling scheme, the transmitter may include in the preamble a full designation of the channel, or a full designation of the carrier frequency used for the th hop or for subsequent hops in the preamble.

The receiver needs only two bits of information to determine the channel index, e.g., if the maximum frequency error corresponds to ± 1.5 times the modulation bandwidth.

In a possible variant of the invention, the information about the initial channel is not explicitly modulated anywhere in the preamble or message, but it is implicitly signaled by the selection of the th hop frequency, the hop sequence is defined in terms of relative spacing, the transmitter can freely select the th hop frequency and by this implicitly signal the channel selected for the th hop of the hop sequence, referring to the subband plan of fig. 3, e.g. the 1.5 bit information can be implicitly transmitted with the selection of B0, B1 or B2 for the th hop, the convention can be that B0 is used if the frequency index modulo 3 equals 0, B1 is used if it modulo 3 equals 1, and B2 is used if it modulo 3 equals 2, the receiver can immediately hop the sequence since the subband can be unambiguously detected.

Fig. 5 illustrates a frequency plan that is not partitioned into disjoint subbands. This avoids guard bands between sub-bands and maximizes spectral efficiency. The frequency span is split into contiguous subbands, which are grouped into different sets labeled a/B/C/D to create a disjoint grouping of subbands. For this reason, the frequency distance between the 2 sub-bands (A-A, B-B, C-C or D-D in the figure) of the same packet should be higher than the maximum transmission frequency error of the end point.

The disjoint sub-bands may then allow implicit signaling using th hopped frequency packets may be used for network planning whereby a given gateway will be assigned a given packet and thus the received sub-band will not be ambiguous in an alternative scheme packets may be used to separate traffic according to data rate, again the th hopped sub-band index is not ambiguous because the data rate is signaled in the header of the frame.

Independently of the use of implicit signaling, the grouping of bands into sub-bands can also be used to separate uplink traffic between users with high power reception and users with low power reception for protecting the weakest users. Due to the frequency error there will still be some collisions at the subband edges, but less than in the case of a defined hopping sequence over the entire band.

Conflict prediction and elimination

Since the system is based on frequency hopping, insofar as the hopping sequence is known, the receiver has a predictive possibility if and when two messages will collide. The preamble and sync portions of the frame cannot be predicted in advance, but the data can be post-processed after a short delay. Since the frame is FEC encoded and interleaved over all hops, the receiver should wait for the last hop before de-interleaving and decoding, at which time the collision prediction and cancellation process for the frame of interest is complete. Therefore, the cancellation process does not delay the decoding process.

In the preferred embodiment, the prediction comprises constructing an interference map in the receiving gateway, which predicts the received signal level as a function of time and frequency, the resolution, or granularity, of the map is preferably equal to or better than the length of the symbol on the time axis and may have a resolution equal to samples.

When a signal is detected, the receiving gateway determines its hopping sequence and fills the bins of the interference pattern corresponding to the expected time and frequency of the incoming packets with the signal level measured in the detection step ( power levels per packet).

The term "signal level" is used herein to denote any suitable indicator of strength or power, including but not limited to measured received signal strength, or RSSI measured in dBmW, in dB μ V/m, or in any suitable scale or unit.

The operation is as follows:

1. the receiver instantiates demodulators per detected frame the demodulators generate log-likelihood ratios (LLRs) for each received bit, and possibly quality indicators such as RSSI, SNR, or others.

2. As the detection and demodulator examples progress, they update the interference graph by adding the measured Relative Signal Strength (RSSI) of the signal of interest, such an interference graph shows the possible cross interference of all nodes that transmit at a given time and should have a time granularity of at least symbols length and a frequency granularity better than the modulation bandwidth.

3. the frame reaches the end of demodulation and before de-interleaving and decoding, the LLRs are weighted and possibly cancelled based on an interference map, calculating the interference plus noise ratio (SINR) for each received bit in the case where the interference map shows simultaneous transmissions in time and frequency from two nodes, the LLRs are weighted by a correction factor typically between and zero to account for the fact that the signals in these time and frequency slots are potentially corrupted by interference.

Preferably, the receiver gateway is arranged to measure the signal level of a plurality of packets in a hopping sequence of interference-free radio signals, and if the signal level of the source changes during a change of reception, the signal level of the packet foreseen in the interference map can be adjusted.

In another preferred variant, the receiving gateway should also assess interference from other systems by comparing the received signal with those forecasts from the interference map.

Cooperative reception

The messages transmitted from sensor nodes may be received by more than gateways, and in this case several receivers may cooperate in their decoding, referring to FIG. 1, the message transmitted from S1 is received by both G0 and G1 for each frame, the receiving gateways G0, G1 transmit the necessary information to the server 105 (which may be in separate locations, or in the same place as the gateways ; separate parts or hardware, or simply an instance of a software program).

Multiple sync word (hopping preamble)

ALOHA networks may use special data sequences, conventionally indicated as "sync words", for identifying the start of a transmission or data frame. The sync word may be placed in the preamble or physical header of the frame and its structure is sufficiently known by the receiver to allow detection and word alignment.

The sync word and physical header are particularly important because if they are misinterpreted, the reception of all subsequent data is impaired. Fig. 6 shows the expected error rate in the detection of sync words of a preamble, which is simulated for system self-interference in an ALOHA network with increased load under the following assumptions: sync words are missed when more than 20% of the duration of the sync word is covered by interference. The preamble is not protected by FEC and interleaving intervening at a higher level and the error rate at 20% load is approximately 28%. For comparison, fig. 7 plots the Predicted Error Rate (PER) for the payload of the same network of fig. 6 assuming convolutional FEC (133, 177, k = 7), and 80% of the interference was detected and cancelled and 20% was not before decoding, which is a realistic assumption for weak nodes. We see that the PER at 20% load is about 8%.

In an embodiment, the sync word, and possibly also the physical header, is transmitted in several copies at different frequencies, following the hopping sequence. Each repetition of the sync word is combined with an element of information on the beat index, e.g., a counter, which allows the receiver to align to the beat sequence. Fig. 8 shows a possible implementation of this multiple header transfer. The plot represents an uplink transmission comprising a time succession of several hops having different carrier frequencies and occurring at different ordinates in the plot. Each individual beat, represented by an empty or filled rectangle, comprises a narrowband modulated signal.

The preamble beat 302 contains a plurality of sync words, each combined with at least beat indices, which are repeated in two copies at the beginning and end of each beat in this embodiment.

In a possible implementation, the hop indices are arranged in a sequence with a reduced countdown, the hop labeled with "0" being the last of the preamble hops.

The number and repetition rate of the sync words 320 need not be fixed, but may be dynamically modified by the transmitter based on a weight factor. In a preferred embodiment, the transmitter adapts the sync word repetition number and/or the sync word repetition rate based on the synchronization status, which indicates the frequency error of its local frequency reference.

In another embodiment, the number of sync words repeated and/or the sync word repetition rate is adapted by the transmitter based on an estimate of the transmitter success rate, the transmitter success rate is the probability that a transmission is correctly received by gateways or by multiple gateways, it depends on the network load and on the received signal level seen by the gateways.

The transmitter may estimate the success rate in several ways. The exact method of estimating the success rate is based on the fraction of transmitted frames that should be acknowledged by the network and in fact by the network. Less accurate methods that do not require acknowledgement include estimating the load on the channel by sampling the channel only for signal level, or by attempting to detect sync words from other transmitters, and calculating the success rate from the channel load.

In another embodiment, a transmitter adapts a number of sync word repetitions and/or a sync word repetition rate based on commands received on a network.A network infrastructure, such as a gateway or server, may estimate a success rate for a given transmitter.

The data beats 340 are numbered on the plot to distinguish them, but need not include explicit indices the receiver at this point has determined their frequency and can demodulate them normally the transmission is highly immune to interference since the data is FEC encoded and interleaved.

As discussed above, the hopping sequence involves an series carrier frequency (fig. 4) centered on non-overlapping channels in a possible embodiment, the transmitter node is arranged to offset the carrier frequency of certain preamble hops by a fraction of the bandwidth, as illustrated in fig. 9.

Although this shift appears to increase the likelihood of collisions in frequency, it is in fact useful because the center frequency of the channel typically used by the transmitter is not well defined. The modulation bandwidth is low, about 100 Hz, and in order to detect any sync word, the receiver needs to form a large number of channels, or increase its bandwidth.

Due to the shifting of fig. 9, the receiver can form fewer channels or reduce its bandwidth and still be able to detect at least the portion of the sync word with high probability if the sync word of half is shifted by BW/2, the receiver is confident that the sync word at half will have a lower shift than BW/4.

Importantly, the sequence of integer and fractional steps follows a deterministic rule so that the receiver can apply these deterministic fractional shifts to the nominal sequence of frequency hops and maintain exact tuning, which is within the range of positioning in the sequence of known shift words. As already mentioned, in the example of fig. 9, the even indexed sync word is shifted down by BW/2. Other deterministic rules may be used for the same effect.

Among these choices, the present invention may use the same sync word in all preamble beats 320, or also use different sync words for each beat, according to a predetermined known sequence, hi the latter case, the receiver may determine the location of the beat from the beat index and/or from the sync word itself.

The preamble beat 320 may convey additional information to the receiver in addition to the synchronization word. For example, they may include additional information specifying the data rate of the data portion, and/or the following indication: which specifies a hopping sequence among a plurality of possible hopping sequences. These elements of information may be encoded in any suitable manner.

Hopping sequence

As already stated, frequency errors in the transmitter node make it difficult to define the hopping sequence. Preferably, for the hopping preamble, and possibly for the hopping sequence at the beginning of the data part, should be such that it can only be identified from the index and the current frequency. In addition, the hopping sequence determination should not be affected by such an offset, since the transmitted frequency may show higher errors than several times the channel bandwidth.

Given a contiguous set of potential frequencies, an advantageous option is that the transmitter is free to select the initial frequency in terms of relative spacing, which means that in a hop sequence, the frequency of a given hop can be derived from the previous frequencies by shifting, followed by a modulo operation, to keep it within the intended band limits, as already mentioned.

As far as the hopping preamble is concerned, all frequencies are relative to the th data hopping center frequency, which we denote as f _ data0.

We define the minimum jitter step as h step. To simplify the description and the figures, we assume that h _ step is equal to the modulation bandwidth BW (e.g. measured at a 6dB cut-off). However, this is not a necessary requirement. The transmitted signal will be between f _ min-bw/2-f _ error _ max and f _ max + bw/2 + f _ error _ max (with a 6dB cutoff).

We denote by N the number of channels available for hopping from the viewpoint of the transmitter, N = floor (f _ max-f _ min)/(h _ step + 1).

From the receiver's point of view, the total set of center frequencies that should be scanned spans from f _ min-f _ error _ max to f _ max + f _ error _ max, where f _ error _ max denotes the maximum frequency error for the transmitter, which is a consequence of its crystal oscillator error.

To simplify notation, we force f _ data _0 to be such that f _ data _0 = f _ idx _ data _0 x h _ step + f _ min, where f _ idx _ data _0 is an integer between 0 and N.

When defining several beat sequences, we label the identity of the beat sequence as hop _ seq _ idx. Then, we can describe the frequency versus index and hop _ seq _ idx. Remember that the index is a beating preamble index, which counts down:

1. if index =0, f _ idx _ preamb (index) = mod (f _ idx _ data _0 + hop _ delta (0, hop _ seq _ idx), N +1)

2. If index >0, f _ idx _ preamb (index) = mod (f _ idx _ preamb (index-1) + hop _ delta (index, hop _ seq _ idx), N + 1).

The number of levels hop delta should be chosen to minimize repeated collisions, especially in preamble hops, if two transmitters happen to collide on hops of a hop preamble they should collide as little as possible on the other hops ideally they should not collide on the other hops if they would collide on a given hop.

1. Hop deltas equal to ± 1 should be avoided. Algebraically: for any index and any hop _ seq _ idx, the inequality abs (hop _ delta (index, hop _ seq _ idx)) > 1 should be maintained. This is because the center frequency is not well defined due to frequency error, so with an offset of only 1, a collision can occur on 2 consecutive hops, as shown in fig. 10. The exact rule, if BW ≠ h _ step, then: abs (hop _ delta (index, hop _ seq _ idx)). h _ step > BW + abs (hop _ offset).

2. The value of hop delta should be different, otherwise must have multiple collisions occurring once collisions occur, which is illustrated in FIG. 12.

3. Due to frequency error, the difference between the jitter increments should be at least 2. This is shown in fig. 11, with jitter increments of 2 and 3.

4. For any index and any hop _ seq _ idx, hop _ delta (index) should be different from any sum of consecutive hop increments. This is illustrated in fig. 13. The exact rule would be the sum of successive beats plus or minus 1, but according to rule 3, the difference is in any case a multiple of 2.

To satisfy the first 3 rules, the hop delta sequence can simply be an odd or even integer. To comply also with rule 4, assume that the maximum number of sync words is 4, 4 different sequences within the hop preamble, these may be 2,4,8,10, 14,16,20,22, 2, -4, -8, -10, 14, -16, -20, -22.

The simulations have shown that: the above rules significantly improve the probability of correctly detecting the preamble and reconstructing the hopping sequence despite frequency errors in the transmitter.

Different hopping sequence options

Preferably, the pseudo-random hopping sequence is bijective from hops to another hops, the frequency of each hop being deterministically derivable from the frequency of the preceding .

Due to the above bijective correspondence, the receiver only needs to detect hops to reconstruct the hopping sequence by itself.

The problem with the -like pseudo-random sequences is that they require exact knowledge of the channel.

The transmitter has a stable frequency reference, such as a TCXO, and synchronizes its frequency from the downlink frame or beacon in this case accuracy may be better than half or even of the modulation bandwidth.

The transmitter has a less stable frequency but does synchronize its frequency from the downlink frame or beacon. In this case, the error may be up to 1PPM, which leaves less than 1KHz in either the 868MHz or 915MHz ISM band. Then as side information, the transmitter can signal the frequency at which it is aimed to transmit, as described here within the hop preamble or by using a different approach. In this way, the receiver knows the frequency error of the transmitter and can derive the hopping sequence. Due to the limited frequency error, only 2 or 3 bits are sufficient as side information: the error is below +/-1.5 channels and so what is important is mod (f _ idx _ preamb, 4). The hop offset can be adjusted so that the maximum frequency error in this case is limited to +/-1.5 channels, or other values, in order to limit the side information needed.

Preferably, the hop sequence is unambiguously derivable from the nominal carrier frequency or channel index of the hops from its th hop or equivalently at the determined position in the detection preamble, possibly with additional side information, although possibly several times larger frequency error than the channel bandwidth.

To address this problem, the present invention relies on a sequence of hops that are defined in terms of relative shifts followed by modulo operations for hops whose frequencies are close to the boundaries of a band or sub-band, the receiving gateway may not be able to know whether the transmitters have judged that they fall outside the band limits and whether modulo operations have been applied.

In fact, the transmitter node will synthesize the frequency based on its own frequency reference, which is affected by the error f _ err, and depending on this unknown quantity, it may not be possible to apply the modulo operation to the beat 232, or to the beat 231 if the error is inverted in sign. There is therefore an ambiguity in that some beat frequencies can be synthesized at two different frequencies depending on whether the delta results in a frequency that is judged by the transmitter to be outside the subband limit.

To cope with this ambiguity, the receiver may open a second listening channel for the beat expected to fall close to the boundary, e.g., when a specified offset is applied to the beat 234, the receiving gateway finds that the next beats 237 will fall outside f _ min, and should then apply a modulo operation and listen for the next beats at the location marked 236.

In this way, there are permutations per starting frequency.

Preferably, the sensor node signals the beat sequence implicitly by the start frequency (which we also refer to as th data beat frequency, or reference frequency).

If index =0, f _ idx _ data (index) = f _ idx _ data _0

If index >0, f _ idx _ data (index, hop _ seq _ idx) = hop _ perm [ hop _ seq _ idx ] (f _ idx _ data (index-1))

The permutation where hop _ perm [ hop _ seq _ idx ] is (0, N-1), preferably less cycles.

This can be obtained by a stable frequency reference in the transmitter, as already described, or by implicit or explicit signaling of any jitter, preferably the absolute frequency of th, as synthesized based on the transmitter's own frequency reference.

The value of the absolute frequency may be signalled implicitly or in any other suitable way, implicitly signalled, by the choice of th sub-band used in the transmission, within which the frequency reference in a given mobile node can be limited to drift, since the sub-band determination is from unambiguous.

Hopping sequences for sub-band cases

If the available frequencies are divided into subbands, the transmitted hopping sequence may be specified by a sequence of subbands that is hierarchically superimposed to the channel sequence. Each hop is transmitted in a different sub-band and in a different channel, as indicated by the corresponding sequence.

To simplify notation, we assume that all subbands have the same number of channels equal to N + 1. In this way we keep a single band case notation to mark the frequencies used within the sub-band.

The subband index is labeled sb _ idx _ data, which depends on the hop number index, and the hop sequence number hop _ seq _ idx for a given hop sequence number, the hop sequence depends only on sb _ idx _ data _0 and f _ idx _ data _0, i.e. the frequency of the th hop.

The hopping sequence is then defined as a set of sb _ idx _ data (index, hop _ seq _ idx, sb _ idx _ data _0, f _ idx _ data _0) and f _ idx _ data (index, hop _ seq _ idx, sb _ idx _ data _0, f _ idx _ data _0)

The organization of the available frequencies as disjoint sub-bands provides the advantage of being independent of hopping sequences, better diversity against multipath, better interference mitigation, better coexistence and as a result better with procedure .

We can use sb _ idx _ data _0 to signal both the beat sequence and side information, such as the LSB of f _ idx _ data _ 0. for example, when there are 40 subbands, 2 bits can signal the LSB of f _ idx _ data _0 and then 10 beat sequences can be signaled from sb _ idx _ data _ 0. of course, still more beat sequences can be implicitly signaled by f _ idx _ data _ 0. We can also signal beat sequences from sb _ idx _ data _0 only and make the frequency error unknown, but then the received would have to be replicated at the edges of the band-open receive window (replicated in frequency), such as in FIG. 14.

For unsynchronized nodes, the space for f _ idx _ data is reduced to ensure that the transmission always occurs in authorized bands. Signaling the LSB merely helps to provide a better hopping sequence.

Assuming that a single set of hops is used (the hopping sequence is determined by only the th hop), the interference mitigation is slightly better in the sub-band case.

Hopping sequences for sub-band cases with perfect frequency synchronization

This case assumes that all sensor nodes have better frequency synchronization than ¼ for the modulation bandwidth, so that there is no ambiguity in the intended frequency.

Here, the th beat frequency, defined by the { f _ idx _ data _0, sb _ idx _ data _0} pairs, is still used to index the beat sequence, potentially with other indices.

It is part of the prior art for generating pseudo-random sequences that depend on f _ idx _ data _0, sb _ idx _ data _0, hop _ seq _ idx. there should be for subband indexes and for frequency indexes examples for this use a "pseudo-random binary sequence" generator with a state size higher than or equal to the total number of bits required to represent f _ idx _ data _0, sb _ idx _ data _0, index, hop _ seq _ idx.

Hopping sequences for sub-band cases with mixed frequency synchronization capability

This situation is more realistic, since not all sensor nodes will be equipped with a good time/frequency reference. Even with prior synchronization, the sensor nodes may exhibit significant frequency drift because beacon/synchronization frames are not very frequent in the LPWAN.

Then we need to have both populations co-exist. It is preferable to give higher jitter space to nodes that are better synchronized. Of course, it is also possible to reduce the space for all nodes so they become identical.

The th hop frequency for loosely synchronized nodes is limited and not allowed to be closer to the band limit than f _ err _ max. the transmitting node still signals the exact frequency through the sub-band used by . loosely synchronized nodes may use only N-P intermediate channels (hop 240), while well synchronized nodes may use any of the N available channels (hop 245). please note that for each hop, the sub-band index changes, but this uses a standard pseudo-random hop sequence.

The signaling of which hop set to use is again from the th frequency index, frequencies in the center can signal a full span hop sequence the set of frequencies should be split into 2 disjoint groups as shown in fig. 16 the th group has only frequencies within the center to ensure that the th hop of a loosely synchronized node is within the allowed boundaries and the second group is the rest.