WO2009027408A2 - High spectral efficiency point-to-point radio system and relevant operating method - Google Patents

Abstract

A Point-to-Point radio system comprises a plurality of transmit antennas and a plurality of receive antennas that can be utilized with a transmission modality selected among a plurality of available transmission modalities, all finalized to transmit and receive digital, single carrier modulated RF signals at frequencies above about 6 GHz. The modulated signal has a single carrier modulation/coding format selectable among a plurality of available modulation/coding formats. Both the antenna transmission modality and the modulation/coding format jointly identify a transmission physical level. The system includes in particular, both at the transmit and at the receive side, suitable signal processing circuits coupled at the input of the transmit antennas and, respectively, at the output of the receive antennas, finalized at the real-time, adaptive change of the transmission physical level as a function of the time-variable quality of the signals received by the multiple receive antennas.

Description

" High spectral efficiency point-to-point radio system and relevant operating method"

The present invention relates to point-to-point radio system according to the preamble of claim 1. In particular, the present invention finds applicability in radio configurations utilizing multiple transmit antennas and multiple receive antennas, that are namely called "MIMO" (Multiple Input Multiple Output) configurations, and realize multiple transmission paths between the transmit and the receive side of a link.

Configurations of this kind are used in particular in the cellular wireless technology, thus operating at RF frequencies much lower than the ones considered here, to realize antenna arrays (the so called "smart antennas"), featuring a higher directivity, or to implement spatial multiplexing, enabling to the increase of the transmit capacity at equal occupied RF bandwidth. MIMO configurations can be symmetrical, of the type MIMO(N₅N), as well as asymmetrical, for instance the MIMO(1, 2) configurations, with single input and two outputs, commonly known also as configurations with receive diversity. In the frame of the present invention, reference will be frequently made to the "Physical Level Adaptivity", also called sometime "Link Adaptivity".

With the term "adaptivity" one refers to the dynamic and automatic process of changing the transmission modalities over the hop, as a consequence of the time variations of the propagation conditions, so as to be "adapted" in the best way to said propagation conditions. More specifically, the Physical Level ("PHY Level") is varied between two or more available options so as to optimally counteract the unwanted effects of the propagation in terms of quality degradation of the received signal.

For point-to-point (PtP) radio systems, the technology of presently available products is based on conventional single input, single output MIMO(1, 1) configurations, with configurable, static modulation, not adaptively changed with the propagation. The use of PHY Level adaptivity in radio systems is up to now limited to cellular Point-to- Multipoint (PMP) systems, whereby a plurality of peripheral terminal stations spread over a certain geographical area (called "Cell") is connected to a center station (the "Master Station"). In general a miscellanea of service typologies is supported, with data services being anyhow predominant.

As an example, in the time-division multiple access modality (TDMA = Time Division Multiple Access), the various peripheral stations are enabled to transmit in different, non- overlapping time intervals in order to maintain mutual orthogonality.

The information to be transmitted by each peripheral station in direction to the Master Station (in the so called "Up-Link" direction, but eventually also in the opposite "Down-Link direction), is consequently segmented in elemental blocks. Such blocks are mapped with suitable modulation formats over packets (or bursts) of signal, transmitted in distinct times. The transmission times are properly adjusted in order do not to have any overlapping at the Master Station receiver, and avoid this way any mutual interference.

The system traffic modality is based on the principle of "capacity assignment on request", resulting in practice in a variable number of bursts transmitted in a given time interval by each peripheral station, so as to realize a variable transmission capacity over time ("time varying throughput").

In these conditions it becomes quite natural to utilize different modulation formats (so different "PHY Levels") for each burst, in order to maximize the overall cell throughput. On the basis of the receive signal quality, one can this way "match" in the best possible way the transmission requirements of the various peripherals with the propagation situations of the various links, situations that can be even very much different from link to link, not only because of the different hop lengths, but also because of the different propagation conditions. Asymmetrical MIMO configurations with Multiple Inputs and a Single Output, MISO(N, 1), have been proposed for the Master Stations of wireless cellular systems to implement multiple elements antenna arrays with sectorial coverage, with the purpose of increasing the overall directivity of the station or of realizing a spatial multiplexing functionality ("smart antennas").

Solution of this type, combined to the modulation format adaptivity, are for instance suggested in the frame of the "WiMax" technology by the relevant IEEE Standardization (refer to IEEE-802- 16).

Next generation PtP products are announced in the literature that will make use of the

"Modulation Adaptivity", that is that type of adaptivity limited to the change of the Modulation format only, without any variation to the transmission modality. MIMO configurations for PtP radio systems are described in the Patent Applications EP 1 229

670, with reference to an individual link, and in the Patent Application EP 1 448 006, with reference to wireless networks. However these systems do utilize a static (non- adaptive) PHY

Level.

The Document WO 2007/069071 describes PMP and PtP links based on MIMO configurations, also operated in an adaptivity regime. However the concept is applied to OFDM modulation formats because finalized to lower frequency bands (<6GHz) in NLoS (Non Line of Sight) propagation environments and high incorrelation between the various propagation paths due to the antennas multiplicity.

The solution suggested in said Document WO 2007/069071 turns out to be not at all suited nor applicable to Single Carrier modulation formats, the only ones functional to high frequency

(>6GHz) PtP MW Radio Systems, which will be in turn operated in LoS (Line of Sight) propagation environments, with high correlation between the various propagation paths due to the antennas multiplicity.

On the other hand, the use of Single Carrier formats for the high frequency MW bands is motivated by fundamental technical reasons, such as the better PAR (Peak- to- Average Ratio) between peak and average signal power for equal emission templates and equal bandwidth efficiency (ratio between useful and total number of tones for the OFDM formats and ratio between symbol frequency and total FR bandwidth for Single Carrier formats).

The worse PAR makes OFDM formats not at all convenient for high frequency MW bands where HPAs with high saturation power are critical and anyhow expensive.

In addition, the high spectral purity (low phase noise) required by OFDM formats are not economically viable for high frequencies (>6GHz), and technically not available for very high frequencies (23 to 38GHz).

Furthermore the NLoS feature offered by the OFDM formats turns out to be completely useless at high MW frequencies (>6GHz), because the propagation has to be LoS anyhow, to avoid dramatic signal losses due to shadowing. For all the above reasons PtP systems based on OFDM as the ones mentioned before, are unacceptable for the high frequency MW bands (>6GHz).

As a consequence, it turns out that the spectral efficiency boost offered by the present invention becomes very important for high frequency (>6GHz), LoS, PtP, MW radio systems, based on Single Carrier Modulation formats. The objectives of the present invention are twofold, namely:

- to introduce a PtP link architecture able to offer a very significant boost in spectral efficiency for high frequency (>6GHz), Single Carrier, LoS links, and

- to propose an innovative operating method for such links, able to overcome technical problems and difficulties of the state-of-the-art technology. The first objective is reached by a PtP radio system in accordance to Claim 1.

The second objective is furthermore reached by an operating method in accordance with Claim

13.

The additional characteristics and advantages of the PtP radio system and relevant operating method according to the present invention will be highlighted in the description that follows and will result from the preferred embodiment also presented in the following, this last provided only as an indicative and non limiting example of possible system implementation.

The description makes reference to the annexed Figures, in which:

Figure 1 shows the vectorial relationship between the various signals in a MIMO(2,2) configuration, Figure 2 shows a block diagram of a MIMO(2,2) configuration,

Figure 3 shows a principle block diagram of a Combiner/Canceler, at the receive side,

Figure 4 shows the vectorial relationship between signals in a lxMIMO(2,2) configuration, Figure 5 shows the vectorial relationship between signals in a 2xMIMO(2,2) configuration,

Figure 6 shows the vectorial relationship between signals in the IxMIMO(1, 2) configuration, alternative to the configuration of Figure 4,

Figure 7 shows an example of signal segmentation for the two transmit antennas, for modulation formats up to 64QAM, and in particular for formats carrying 2, 4 and 6 bits per QAM symbol,

Figure 8 shows a schematics relevant to the control of PHY Level switchovers in the MIMO

(2,2) configuration,

Figure 9 shows a comparison between adaptive PtP systems using various PHY Level profiles,

Figure 10 and 11 show the exemplary performance of some system configurations according to the present invention,

Figure 12 shows exemplary antenna separations with one of the system configurations of Figure

10,

Figure 13 shows a general block diagram of a PtP radio system according to the invention,

Figure 14 shows an example of PHY Level adaptivity process flow, given a certain PHY Level profile.

In the description that follows frequent use is made of the term "adaptivity" to indicate the process by which the transmission modalities over the link are dynamically and automatically adjusted, as a consequence of and in response to the variations in the time of the propagation conditions experienced by the link. More specifically, at constant occupied RF bandwidth, the PHY Level is varied among two or more available PHY Levels, and "adapted" in real time to the variable propagation conditions in order to counteract in the best possible way the signal quality degradations caused by said variable propagation conditions.

It is important to notice that here (and in all the following description) the term "PHY Level" indicates in a more general sense the combination of all the operations performed on the signals to be transmitted, jointly defined in particular by: the selection of a specific Modulation/Coding format ("Modulation Mode") among two or more available formats, and concurrently the selection of a specific transmission modality ("Transmission Mode") among at least two available Modes

This is unlike the meaning and the use of the same term "adaptivity" that is made for conventional PMP radio products or even for next generation PtP products near to appear on the market, where adaptivity is meant and used limited only to the selection of the Modulation

Mode.

The Transmission Mode also sets the multiplexing factor m of the signal.

For a certain RF channel bandwidth BW, a generic PHY Level allows for the transmission of a net capacity T (also indicated as "Payload" or "Net Troughput"), related to BW by the following simple relationship:

T (Mbit/sec) = m*η*BW/(l+p)

Where : m is the signal multiplexing factor BW is the RF occupied bandwidth (channel bandwidth) η is the net spectral efficiency (due to both the QAM format and the frame overhead) p is the roll-off factor of the signal shaping (in practice 0,2< p <0,5)

With MIMO(2,2) and configurations using a single polarization, the multiplexing factor m can be 1 or 2, depending on the selected Transmission Mode. In case of polarization reuse (also possible), the spectral efficiency is further doubled and m can be 2 or 4

The spectral efficiency becomes the higher, the higher is the complexity of the QAM format and the lower is the transmission overhead, this last also determined by the FEC capability in terms of multiple error correction. Obviously, also the performance of the PHY Level, represented by the "threshold" value α of the signal-to-noise power ratio (SfN) required for a given, still acceptable error probability (BER =

Bit Error Rate), changes with the complexity of the PHY Level. In other words two parameters (η, α) can be associated to any PHY Level, which are mutually interrelated, in the sense that to realize more performing η values, one has also to rely on higher threshold values α.

Let's indicate with PHY_MIN the minimum PHY Level, representing the most robust level (minimum α) and with PHY_MAX the maximum PHY Level, representing the most efficient one (maximum η).

If, by the variations of the propagation conditions, the rain attenuation over the hop becomes higher, one switches down to a less complex PHY Level, thus implying a lower α value, with the result of counteracting the increasing BER degradation. At constant bandwidth BW, this will cause a reduction of the spectral efficiency η and a corresponding reduction of the instantaneous throughput T.

In other words, the PHY Level adaptivity also implies the time variation of the instantaneous throughput carried by the system. Consequently this solution can only be adopted for the transmission of digital signals that can tolerate the short-term variation of the transmission rate, so typically for the transmission of payloads containing, at least for a significant fraction, "best effort" data services not linked to a real-time transmission, be they of the so called "Cell oriented" type (for instance ATM mapped on IMA-NxEl physical interfaces), or of the so called "Packet oriented" type (for instance IP over Ethernet 10/100 Base T physical interfaces, as typically found in the IP access network). The PtP radio system according to the invention operates at high MW frequency bands (typically >6GHz and more commonly >13GHz), benefitting of the quite large frequency blocks offered by the relevant standardized channel plans and of the limited hop lengths to bridge (few kilometres in urban/suburban environments and typically <15/20Km in rural environments). The available RF channel bandwidth is generally a multiple of 3,5 MHz for the ETSI standards, BW = [3,5 - 7 - 14 - 28] and a multiple of 2,5 MHz for the ANSI standards, BW = [2,5 - 5 - 10 - 20]. The links we consider are in full LoS, with highly directive antennas, and negligible dispersive effects within the channel band BW. This is due to the non-selective (flat) in-band response of the supplementary attenuation, predominantly due to the following two propagation phenomena:

1) Multipath, with echo delays much smaller than 1/BW, the inverse of the channel bandwidth, because of the limited values of both the hop lengths and the channel bandwidth themselves. This effect can be considered predominant for the lower RF bands, from 6GHz to 13GHz

2) Rain, that can be in turn considered to predominate for the higher RF bands, greater than 13GHz

With these assumptions let's consider a configuration with a plurality of transmit antennas and a plurality of receive antennas, briefly a MIM0(N,M) configuration, and, in particular, a MIMO(2,2) configuration as in the example sketched in FIGURE 1.

Given the flatness of the channel response and since of course the unequality s <« L holds between the antenna separation s and the hop length L, the direct and cross paths between the transmit and the receive side antennas can be considered perfectly correlated. With signals Sl and S2 transmitted by the two transmit antennas, Tl and T2 respectively, the two receive antennas will receive the signals Rl and R2 given by :

Rl = wii*Sl + wi2*S2*e^JΦ21 ~ w_o*(Sl+ S2*e^JΦo) [1]

R2 = w₂₂*S2 + W2i*Sl*e^JΦ12 ~ w_o*(Sl*e^JΦ° +S2)

The high correlation between the paths implies in fact that the coefficients w _xx are all approximately equal to a common value (wo) and also that it holds Φ21 ~ Φ₁₂ ~ Φo. By identifying in the term e ^jΦo the electrical phase difference between cross and direct paths, such difference is essentially determined by the difference in the physical lengths ΔL. It holds: Φo/360 = ΔL/λ with ΔL = sqrt (L² + s²) - L ~ s²/2L and finally :

Φo (degrees) « 360*(s²/2Lλ) "Characteristic Parameters" of the PHY Level : By changing PHY Levels, the modulation format switchovers are assumed to be performed at constant static peak power of the QAM constellations, so as to operate the transmit HPAs always in the same condition of linearity With this assumption the system gain SG is defined as the difference in dB between the static peak power PPTX (dBm) transmitted by each antenna and the receive threshold THRX (dBm) for the wanted BER objective:

SG (dB) = PPTX - THRX

THRX (dBm) = -114 + NF + 10*Log(B_N ) + α where:

NF(dB) is the receive noise figure

B_N (MHZ) is the equivalent receive noise bandwidth α is the threshold (SfN) ratio between the static peak power of the received signal and the thermal noise, for the wanted BER objective B_N is obviously related to the channel bandwidth BW. For QAM formats, raised cosine shaping and optimum subdivision of filtering between transmit and receive side, one has:

B_N = Fs = BW/(l+p) where Fs is the symbol frequency and p the roll-off factor of the selected channel shaping (with practical values in the range 0,2< p <0,5). To be noted that, with the relationships above, the System Gain difference between two PHY

Levels operated at equal PPTX and BW values, and using the same receivers (same NF and same

B_N ) turns out to be:

ΔSG (dB) = - Δα that is equal to the difference between the relevant values of α, but with sign inverted. The threshold for a given PHY Level is simply given by the condition of having the system gain SG (offered by that PHY Level) equal to the total attenuation present over the link. In other words it holds for the threshold :

SG = (ASL + As) where: ASL (dB) is the free space attenuation (inclusive of the gain of transmit antennas Tl and

T2 and of receive antennas Rl and R2)

As (dB) is the supplementary attenuation due to the propagation. For a given hop length L, ASL and the remaining parameters (PPTX, B_N, NF) are set, so that the condition for the switchover from a PHY Level with threshold α down to a lower (more robust)

PHY Level [or the condition for the link threshold if the PHY Level is already the minimum one] is given by the relationship: As < SG - ASL = costant - α = KK - α that namely provides the maximum value of As that can be tolerated for a given PHY Level [or for the link threshold].

The value of the constant KK depends on all the "fixed parameters" of the link, that is all the parameters which do not change with the PHY Level switchover (L, PPTX, NF, B_N, BW, F_S). The higher is the tolerable attenuation As (or the lower is the α value of the PHY_MIN), the lower is also the outage probability POUT, corresponding to the probability of having a BER worse than the objective.

From all the above it turns out that a proper characterization of the link performance offered by a generic PHY Level is conveniently described by the pair of "characteristic parameters" [η , RSG], where η (bit/Hz) = net spectral efficiency

RSG (dB) = Relative System Gain = - α being RSG relative to an arbitrary reference, common to all PHY Levels (e.g.: RSG = - CC , with respect to the arbitrary reference KK=O). Selection of the PHY Levels : The configuration MIMO(2,2) can be used with two different modalities (Transmission Modes), when adopting PHY Level adaptivity.

Modality "Ix" - The quadruple antenna arrangement is not used for spatial multiplexing but to realize diversity configurations only, that can be in principle either of order-2 [simple RX diversity, by switching off one transmitter and realizing actually a SIMO(1, 2) operation] or of order-4 [combined TX and RX diversity, by fully exploiting the installed (2,2) antenna basis]. In both cases the payload multiplexing factor is m=l. Obviously the simple RX diversity option has little interest once a (2,2) antenna configuration is anyhow available. Modality "2x" - The quadruple antenna arrangement is used to maximize the throughput with an order-2 spatial multiplexing. The payload multiplexing factor is m=2, and the throughput is doubled.

The purpose of the PHY Level adaptivity is essentially to introduce an additional degree of freedom in the system, so as to independently select the values of the two "effective" parameters which govern the overall link performance:

- the RSG of the lowest PHY Level (PHY_MIN), responsible for the link POUT

- the η value of the highest PHY Level (PHY_MAX), which is in turn the main responsible for the average spectral efficiency of the link. In practice the PHY Level profile will be selected with the following criteria:

- PHY_MIN : a lower η is allowed, anyhow able to support the minimum required "highest availability" throughput Tmin, whose outage probability has to satisfy the established POUT objective. This leads to the choice of a robust Modulation/Coding Format (e.g. 4QAM) associated to a "Ix" Transmission Mode in order to jointly minimize the α value, and consequently the POUT.

- PHY_MAX (and higher PHY Levels) : the choice goes to more complex Modulation/Coding Formats associated to "2x" Transmission Modes, in order to maximize the average spectral efficiency.

Looking at the link performance the block diagram of one direction of the link is presented in FIGURE 2, the other direction being perfectly reciprocal.

The radio system comprises a PHY Level Transmit Processor 1 (in charge of transmit segmentation, Service insertion and FEC encoding, mapping and dual modulation), two RF transmission modules 4 and 5 (each including an IF /RF up-conversion module and a power amplifier) driving the relevant transmit antennas Tl and T2, two RF receive modules 6 and 7, connected to the relevant receive antennas Rl and R2 (each module including a low noise RF amplifier, and a RF /IF down-conversion module), a Combiner/Canceler 8, and a PHY Level receive Processor 11 (in charge of dual demodulation, demapping, FEC decoding and service extraction and data serialization).

The values of α for the various PHY Levels are of course meant at the input of the receivers

(section A-A of FIGURE 2). With respect to the corresponding values β at the input of the demodulators one has: α (dB) = β - GcoMB where G_COMB is the (SfN) gain provided by the Combiner/Canceler block 8.

We will have consequently :

RSG(dB) = - α = GCOMB - β

By using QAM Modulation formats it is convenient to introduce a compact notation "KxNS(N,M)" to indicate a generic PHY Level, where:

K stays for the MIMO Transmission Modality (K=I for modalities "Ix" and K=2 for the modality "2x")

NS = number of states of the QAM constellation

(N,M) = number of used inputs and, respectively, outputs of the MIMO configuration. For instance the notation "1x4(1,2)" will indicate a PHY Level using 4QAM with simple RX diversity, for the Transmission modality "Ix".

In turn the notation "2x64(2,2)" will indicate a PHY Level using 64QAM with order-2 spatial multiplexing, for the Transmission modality "2x".

The Combiner/Canceler structure can be represented by the principle block diagram of FIGURE 3, valid for all MIMO Transmission modalities.

In practice each of the received signals Rl and R2 is multiplied by a proper coefficient (C₁₂ and

C₂₁ , respectively), shifted by a proper phase γ and summed to the cross signal R2 and Rl, respectively.

Starting from the expressions of the desired signals Ul and U2 at the Combiner/Canceler outputs, one easily obtains the optimum values of the parameters for each of the various PHY

Levels, as presented in TABLE 1.

These are the mathematical (ideal) values required for the various MIMO modalities At this point it is worth recalling that the PHY Level IxNS(1, 2) has in practice little relevance once a (2,2) antenna configuration is anyhow available to allow for the "2x" modality. It will be considered in the following only for generality reasons.

The by far preferable "Ix" modality is in fact lxNS(2,2) which provides a System Gain advantage of about 6 dB over the IxNS(1, 2) alternative, having only the modest drawback of requiring a return channel to optimize the TX diversity.

TABLE 1 - Ideal parameters for the Combiner/Canceler

PHY Level |MIMO | Parameters in the |G_COMB(dB) I Note Type I modality | Combiner/Canceler

Ix NS (1,2) "Ix" TX Signals (Sl, 0) 20 Log (2/V2) = 3 (1)

RX Signals (Ul Δ Sl, XX)

γ ^ (- Φo)

Ix NS (2,2) "Ix" TX Signals (Sl, Sl) (2)

RX Signals (Ul Δ Sl, XX) 20Log[(4/V2)*cos(Φo/2)]

γ -» 0

2x NS (2,2) "2x" TX Signals (Sl, Sl) 20 Log

RX Signals (U I Δ S2 , U2 [(2/V2)*sin(Φo)]

A Sl)

γ -i ^► (180 - Φo) (degrees)

(1) Legenda : Δ = proportional to (- -) XX = not used

(2) Requires the remote control of the relative emission phase of transmitters Tl and T2. Note that for both "Ix" alternatives the U2 output is not used and that a remote control of the transmit phase difference between Tl and T2 is required in the lxNS(2,2) case.

Controls and algorithms: It is worth to repeat that the PtP Radio System according to the present invention is characterized by the combination of all the following features: -it is based on MIMO configurations [in particular on the MIMO(2,2) one]

-it uses Single Carrier modulation formats

-it exploits PHY Level Adaptivity, being the PHY Level jointly defined by a Modulation Format and a Transmission Modality.

In the following a method and the relevant algorithms are described, able to support the target system functionalities, well aware of the fact that, if alternative methods can possibly be found or proposed, they would nevertheless have to guarantee equivalent functionalities.

The modality 2xNS(2,2) requires a good coherence of the receive MW LOs to optimize the cancellation.

The modality lxNS(2,2) in turn requires good phase coherence in the transmit side to optimize the TX diversity.

Since MW integrated front-ends use typically a single LO for both IF/RF and RF/IF conversions, we will assume that LOs with good phase coherence on the two antennas be available both in the transmit and in the receive side for the system implementation.

The first functionality required at the receive side is the fine level adjustment of Rl and R2 (e.g. through individual AGC stages AGCl and AGC2, respectively), to obtain equal average powers and compensate for possible gain differences in all the blocks responsible to convert said signals down to baseband (to DC) (imperfect antenna pointings, resulting in slightly different antenna gains on the two ways, slightly different gains of the RF/IF conversion stages, and similar).

Then the other variables one needs to precisely control can be derived from the analysis that follows, carried out for each of the various MIMO modalities. The analysis of the IxNS(1, 2) modality is given just for completeness, but the modality is in practice of little interest, as already pointed out. Modality lxNS(2,2) of combined (order-4) TX and RX diversity (FIGURE 4) This modality requires that the relative phase difference of the signals radiated by transmitters Tl and T2 be controlled, in order to achieve the maximum level for both the signals Rl and R2. To this end a phase shift α is inserted on one transmission way, to compensate for any phase difference arising between the two signals Tl and T2. The phase shifter by α has to be controlled in the reverse direction from the remote receive side through the return channel inserted in the system telemetry, with the target to bring to zero the difference between the modules of Rl and R2. In other words, if a relative phase difference β arises between Tl and T2 [difference due for instance to geometrical errors in the rectangular shape of the (2,2) antenna arrangement, latency differences, slight misalignments or tuning spreads in the two TX front- ends, and similar], α has to be driven by the control algorithm to reach the value α = - β for the necessary compensation. The situation is presented in FIGURE 4. As resulting from the simple relationships that follow, the loop control signal C is obtained by the difference of the modules of Rl and R2, and conventional In-Phase-Combination (IPC) is used at the receive side to compensate for possible latency differences or misalignments of the RX blocks.

Rl = wo*Sl + woSle^l(α44>o) R2 = wo*Sle^l(φo) +wo* Sl e>^α

|U1| = |(4wo)Sl*cos(Φo/2)cos(α/2)| maximum for α^O, at equal Φo C = |R11 - |R2| = - (4wo)S 1 *sen(Φo/2)sen(α/2)

Modality 2xNS(2.2) of order-2 spatial multiplexing (FIGURE 5)

In this modality and in the assumption that the remote loop on the transmit side has already reached its convergent point α ^ 0, the signals resulting from the Canceler, are: Ul = Rl + R2*e^Jγ21 = wo*Sl*[l+ e**^*^] +wo*S2* [e>^φ0 + e¹⁷²¹] U2 = Rl*e^Jγ12 + R2 = wo*S2*[l+ e>^(φo+yl2)] +wo*Sl* [e^⁰ + d^yl2] The situation is presented in FIGURE 5 and the convergence condition is represented by γl2, γ21-* (π - Φo) .

A conventional MMSE recursive method ("stochastic gradient algorithm"), well known to the experts in the art, can be used to control and update the values of the coefficients γ by the minimization of the decision errors εi = <U1> - Ul ed ε₂ = <U2> - U2 where the notation < > namely stays for "estimated data" , out of the decision block.

The following recursive algorithms yield the updated values of the coefficients: γl2 ₍κ₊i) = γl2 κ + μ* ε₂κ*Rlκ γ21 ₍κ₊i) = γ21 κ + μ* εiκ*R2κ where μ is the process step-size. Note that relatively low error probabilities (lower than about 10%) are required in order for the algorithms to properly converge. This implies enabling the MMSE controls starting from values of γ already near to the convergence, values that a self-training routine running at the initial system start-up can easily identify and store in the configuration SW.

At the convergence that yields : γl2 , γ21-» (π - Φo) , the resulting signal values out of the canceler are :

|U1| = 2wo*S2*sen (Φo)

|U2| = 2wo*Sl*sen (Φo)

Modality IxNS(1.2) of simple RX diversity (FIGURE6)

Although of little interest, this alternative "Ix" modality is briefly commented just for generality reasons.

To realize maximum power (in-phase) combination, R2 has to be phase-shifted by a proper angle γ and then summed to Rl. For this modality S2 = 0, hence γ ^ (-Φo) if then a sum of the two signals is actually made.

By performing instead the difference of the two signals, clearly one has to converge to γ ^ (π -

Φo). Note that this situation turns out to be much more convenient, because yielding a convergence value identical to the one required in the modality 2xNS(2,2). With reference to FIGURE 6 one obtains : |U1| = | wo* [Sl- Sle^j(φo+γ)] | namely maximized for γ ^ (π - Φo), that yields: |U1| = 2wo* Sl

Frame structure : The features required for correct operation during PHY Level switchovers are essentially two :

- "Hitless" switchover : Phy Level switchovers shall occur instantaneously without transients implying possible bursts of errors, so they shall be "hitless". To this regard changes of the modulation format are irrelevant, whilst changes to the Transmission Mode have to be properly managed.

- Continuity of the controls : the various control loops and algorithms updating the values of the various involved variables shall not suffer of any discontinuity at PHY Level switchovers. One has to realize the continuity over time of the various control loops, independently of the selected PHY Level, and well decoupled from the switching process itself. The control loops to be managed are the following: - AGC1. AGC2

- Control of α and of the IPC (In Phase Combiner) - Control of the cancellation parameters γl2 and γ21

To this end, the proprietary method proposed here consists in making available suitable service fields in the data segmentation, reserved to proper unique words (UW = Unique word), always present in the frame, allowing for the computation of the error signals just during said UWs for the various involved control loops. Being the UWs permanently present in the frame, one achieves this way the complete independence from the PHY Level switchover. Moreover being the inserted UWs of two kinds, the first one dedicated to support the "Ix" modality and the second one to support the "2x" modality, one also realizes the complete independence (orthogonality) of the various loops and their continuity over time, independently of the actual Transmission Mode in use and without any discontinuity arising from the switchovers. We will refer to this method as to the "Dual UW" one. Two different UWs are consequently made available for each transmit way: - U , identical on the two ways Tl and T2 ,dedicated to the "Ix" modality, in charge of the control of the transmit phase CC , and of the IPC, besides of other "housekeeping" functions (e.g. frame synchronisation, link identification)

- A , resulting in two different words Al and A2 on the two ways Tl and T2 respectively, in turn dedicated to the "2x" modality and in charge of the control loops of AGCl, AGC2 and of the Canceler parameters γl 2 and γ21.

Considering then Single Carrier QAM and, for instance, formats having number of states NS = 2^2L (so 4QAM, 16QAM, 64QAM,....), the coding structure that turns out to be convenient, is for instance based on independent Codecs operating in parallel, each on a bit pair (the 4QAM format activates one Codec only, 16QAM activates two Codecs, 64QAM activates three Codecs, ). Considering in addition that the remaining overhead (for FEC error correction, MAC Telemetry and return channel, and for the EOW service channel) can be concentrated in two additional fields (FEC and SRV = Services, respectively), the suggested frame formats on the two ways is shown by FIGURE 7. Different segmentation choices are always possible, but need to support equivalent functionalities.

To be noted that the service field SRV can be provided only for the two most significant bits (4QAM), and left available to the payloads for the less significant bits.

The modulation format for the UWs (both U and A) is here imagined to be 4QAM, but an even more robust 2PSK format could be selected, to increase the probability of correct UW detection, thus making the frame synchronization more reliable.

In case the IxNS(1, 2) mode is adopted as "Ix" alternative despite its poorer System Gain, the Payload transmission would be organized as follows for the two Transmission modalities of the profile:

This case would be an exception to the "Dual UW" method explained before.

In fact no A2 can be transmitted because T2 is off, so that even Al can be omitted. AGCl, AGC2 can be permanently controlled with the U word regardless from the used modality.

When in the "2x" modality, the Canceler can be controlled by a "continuous" MMSE algorithm running also during the payload fields, and the receive IPC is kept updated with the U word.

When in the "Ix" modality viceversa, the receive IPC can be controlled continuously during the payload and the canceler is updated on the U word, thanks to the identical convergent value γ ^■* (π- Φo) for both the "Ix" and the "2x" modalities.

But let's concentrate now on the by far more interesting profile based on the full exploitation of the (2,2) antenna arrangement.

In this case the "Dual UW" method can be properly applied and, besides the Overhead fields treated as previously described, the Payload transmission is organized as follows for the two

Transmission modalities of the profile:

Scrambling with known PRN sequences or doing pre-assigned changes over time of the UWs will allow to avoid systematic words to be transmitted, that would cause undesired discrete components in the radiated spectra. FIGURE 8 presents a detailed scheme of the TX and RX functionalities, for what concerns in particular the management of the PHY Level switchover of this (2,2) profile.

To simplify the drawing, other functions essential to the proper operations of the radio system are not shown, apart from the ones specifically related to the PHY Level switchover.

Omitted functions, by the way well familiar to any expert in the art, include functions and circuits such as

IF/RF and RF/IF Converters and relevant LOs, HPAs and LNAs, IF frequency Synthesizers,

QAM Modulators and Demodulators, IF AGCs (coarse), Cable Interfaces, and similar.

The functions for the TX PHY Level processing (TX segmentation, FEC encoding, UW and service bytes insertion, Modulation and Mapping) are performed in block 101. Block 101 is in charge to prepare the data on the two transmission ways for the successive D/A conversions and IF modulations.

Clearly Block 101 is configured under the external PHY Level selection command provided by the PHY Level Controller, not shown in the Figure.

The T2 way is phase rotated by an angle CC with respect to the Tl way by the Block 102, to properly optimize the TX Diversity. To this end, Block 102 is controlled in remote loop from the

RX side through a proper return channel inserted in the service field of the return frames.

Neglecting the RF functions not shown in the drawing, signals arrive at the receive side, where

Rl and R2 undergoes fine level adjustments to achieve the same power level out of the amplifiers 203 and 204 (AGCl and AGC2). The relevant control signals are obtained by power measurements (I²+Q²), performed at the output of the AfD converters by the power detectors 106 and 107 , enabled only in correspondence of the

A-type UWs (Al and respectively A2), that are specifically dedicated to the Dual Payload Mode.

This operation is controlled by an enabling waveform (S_A) provided by the Synchronizer Block

115, namely in charge of the frame synchronization recovery. The control signal for the phase shifter α in the TX side is provided by the comparator 105 which computes the difference of the modules of Rl and R2 , enabled by the proper enabling waveform (Su), only during the U-type UWs (U on both antennas), that are specifically dedicated to the Single Payload Mode.

The "In-Phase" sum (IPC Block) of Rl and R2 provides the combined output for the lxNS(2,2) Transmission Mode (PHYmin).

The signals are further processed by a (2,2) cancellation structure, performing the sum of the signals Rl and R2 with the respective phase shifted and cross-exchanged versions. The phase shifts γl2 and γ21 provided by the Blocks 108 and 109 are controlled by the MMSE processors of Block 112 that can compute stochastically updated values continuously, even during the payload fields, when the operating PHY Level is 2xNS(2,2), but are enabled only in correspondence of the A-type UWs (Al and respectively A2), when the operating PHY Level is lxNS(2,2).

The step size of these processes is also adaptively changed with the PHY Level, being suitably low to properly average-out the high noise fluctuations present when PHYmin is selected, and conversely suitably high when PHYmax is selected and the noise becomes low. The key for proper operation of the overall structure stays obviously in the quantization size and in the precision of the used arithmetics, and in the correct operation of the Synchronization Block 115.

Block 115 has to provide for correct and reliable receive frame synchronization, utilizing by the way conventional synchronization techniques well known to the expert of the art, in the field of digital multiplexing/demultiplexing techniques or of decoding synchronization techniques for FEC block coding.

Correct synchronization will also depend on the size selected for the UWs (U and A). Selectors 110 and 111 finally select the receive signals and provide them to the decision Blocks 113 and 114 under the external control of the PHY Level Controller with the following correspondence: position 1 = lxNS(2,2) Mode position 2 = IxNS(1, 2) Mode position 3= 2xNS (2,2) Mode To have equal signal amplitudes entering the decisor in both 1 and 2 positions, the signal delivered by the IPC block has to be priory multiplied by w = sen[(π-γ21)/2].

The signals after decision are finally processed by Block 116, also configured under the control of the PHY Level Controller. Block 116 (RX PHY Level processing) includes the functionalities of Demapping, Demodulation, FEC Decoding, Service Extraction and Serialization of received data.

System Performance : As practical example, TABLE 2 presents the characteristics of a possible

PHY Level Profile that can be used with the MIMO(2,2) configuration.

TABLE 2 : Profile example with MIMO(2,2)

The modulation formats include 4QAM, 16QAM and 64QAM, this last reasonably assumed as

"maximum complexity" format.

The modality for PHYmin is the lxNS(2,2) one, based on combined TX and RX diversity.

Practical implementation loss values and a reasonable hysteresis margin (3dB) are also taken into account.

RS (Read Solomon) block coding is assumed with a short block (86 bytes) and a correction capability of up to 3 errors per block. A moderate frame efficiency (0,86) is required to accommodate the overall frame Overhead, estimated in 12 bytes (for the two unique words U and A, the EOW, the MAC telemetry and return channels and for FEC error correction). The uuencoded level of BER = 1E-3 (corresponding to a coded BER ~ 1E-7) is taken as BER threshold for the link.

To be noted that all these choices are only indicative, and that other segmentation and coding solutions can be adopted as well.

A conventional single antenna configuration utilizing adaptive modulation with the same modulation formats would have indicatively the characteristics summarized in TABLE 3.

TABLE 3 : Corresponding profile with a conventional system STND(I,!)

STANDARD (1,1) Profile Frame Efficiency = 0,86

PHY Level Muxiπg (S/N)avg η Impl-loss Peak/avg Margin beta GCOME η RSG (dB) QAM (dB) (dB) (dB) (dB) (dB) (net) (dB)

1x4 (1 ,1 ) 1 ,00 9,54 2,00 1 ,50 0,00 0,00 11 ,04 0,00 1 ,72 -11 ,04 1x16 (1 ,1 ) 1 ,00 16,53 4,00 2,00 2,55 3,00 24,08 0,00 3,44 -24,08 1x64 (1 ,1 ) 1 ,00 22,76 6,00 2,50 3,68 3,00 31 ,94 0,00 5,16 -31 ,94

By representing the two profiles in a (RSG, η) plane, the big advantages offered by the MIMO(2,2)solution are immediately put in evidence, both in terms of spectral efficiency and of available system gain as shown by FIGURE 9.

To be noted that the value of Φo influences the RSG values of the higher PHY Levels, but has practically no effect on the RSG available for the link threshold at PHYmin.

Link quality with PHY Level Adaptivity The quality of a link adopting PHY Level Adaptivity is no longer defined by the simple POUT probability occurring at PHYmin.

Since the higher PHY Levels are also activated for a (by far) predominant time fraction, an additional parameter is needed to express the system capability to provide for throughputs higher than the minimum one. To this end, it is convenient to use a parameter measuring the average spectral efficiency and indicating for instance the amount of its difference with respect to the maximum possible spectral efficiency offered by the PHYmax.

This additional parameter can be called "Throughput Loss" (TL), gives the average relative throughput loss of the link with respect to the maximum one, and is simply defined as: TL = (ηmax - <η>)/ ηmax

Knowing the shape of the distribution of the rain attenuation As over the hop, the average spectral efficiency <η> can be easily calculated as:

<η> = J η(A_s)*pdf(A_s)*d(As) where η(As) represents the "efficiency staircase" offered by the selected PHY Level profile.

For application to cellular backhauling, an excellent "carrier class" quality objective for a MW link using PHY Level Adaptivity with a BER threshold of, say, BER< 1E-6 , can be set at:

POUT ^■*^• 1E-5 TL ^■*^• 1E-3

Advantages offered by the invention: The very significant performance advantages offered by the proposed solution can be best appreciated by comparing on the same hop length a MIMO

(2,2) configuration, according to the present invention, with a conventional single antenna solution STND(1, 1).

For a fair comparison, the STND(1,1) configuration is not based on static modulation, but also uses adaptive modulation, operated over the same PHY Level profile. The basic advantages the invention provides are:

- A significantly higher System Gain at threshold (higher by about 9dB), and

- The duplication of the average spectral efficiency

To better realize the importance of these advantages, it is worth making the following comments:

- The throughput duplication with the conventional technology is only possible by utilizing a second RF channel and a second equipment. Both the HW cost (part of the CAPEX) and the RF licence cost (part of OPEX) would double, while the proposed solution would cost significantly less than two conventional STND(1,1) equipment, because of remarkable synergies offered by a MIMO(2,2) implementation. Moreover the MIMO(2,2) solution would keep the RF bandwidth unchanged. - The required PPTX per antenna would be significantly reduced (by about 9 dB) at equal hop length. This translates in significant cost savings with respect to the two HPAs required to carry the same capacity with the conventional solution. The RF power consumption is reduced, with the associated benefits of bringing down the costs of the DC/DC converters and the operating temperature of the high power RF devices, thus improving the overall HW reliability.

- In alternative to the PPTX reduction one could decide to reduce the antenna diameters [two antennas per terminal for both the MIMO(2,2) solution and for the dual STND(1,1) case], with a resulting significant improvement of the environmental visual impact, being this aspect not at all negligible for urban applications, in particular.

A practical example : To quantify the performance advantages of the proposed solution, we refer here to a practical case in a significant ETSI RF band: 23GHz.

We will compare the MIMO(2,2) and the conventional STND(1,1) solutions, by adopting for both the same 2-state PHY Level profile. In fact 2-state profiles allow for the maximum MAC simplification, without appreciably impairing the system performance.

One can for instance calculate that, at equal outer levels (PHYmax and PHYmin), a 4-state profile would offer only marginal advantages over the corresponding 2-state profile (it would about halve TL, or it would only allow for a marginal 10% reduction of the antenna separation, at equal TL).

The two solutions are consequently compared with the same 2-state profile, as better shown in

TABLE 4.

TABLE 4 : Comparison of solutions using the same 2-state profile

The probability distribution of the rain attenuation As (the only effective propagation component at 23GHz) is provided by ITU-R P.1057-1 "Probability distributions relevant to radio wave propagation modelling".

Being X = (As/Ao,oi% ) the ratio between As and Ao,oi% , the attenuation exceeded only for the 0,01% of the time, (both expressed in dB), one has:

CCDF (As) = 0,01*exp {-14,62 + 0,02326*sqrt[395145,16 - 99011,16*Ln(8,33X)]} valid for latitudes > 30 degrees, with the notation CCDF indicating "Complementary Cumulative Distribution Function" . The following assumptions are then made concerning the main system characteristics.

Segmentation/Coding For BW = 3,5MHz and PHY_MIN = 4QAM : frame duration = 86 bytes, 64Kbps each.

Gross Throughput = 86*64 = 5504 Kb/s and F_s = 2752 KHz (*)

Roll-Off = 100*(3500/2752 - 1) = 27,2%

FEC Coding = Shortened RS (86,74), with triple error correction capability

Threshold BER: Uuencoded = 1E-3, ^■* Coded « 1E-7 Overhead bytes = 12 (6 for FEC error correction, 4 for UWs U and A, 1 for the EOW, 1 for

MAC telemetry and return channels)

RF Parameters

RF Band = 23 GHz , with channel bandwidth BW = 3,5 MHz (F_s = 2,752 MHz)

Zone K (42mm/h) and vertical polarization (K_v = 0,094 , α_v= 1,043) 30 cm Antenna (Gant = 34,5 dB)

NF = 8 dB

PPTX = variabile in the range from 10 to 20 dBm

(to consider that the IdB compression point will be required to be 4 to 5 dB higher)

Quality Objectives POUT = 1E-5

TL = 1E-3 (*) The PHYmin with 4QAM of options A) and B) would carry a capacity equivalent to the presently used products (2xEl in the 3,5MHz band), while options C) and D) would carry two times as much.

The MIMO (2,2) configuration has an additional parameter to select, namely the antenna separation.

As clearly shown by FIGURE 9 (even if referring to a 4-state profile), the variation of the separation s (whose square is proportional to the angle Φo between cross and direct paths) has no appreciable effect on the PHYmin (thus on the POUT), while impacts directly the PHYmax RSG. The separation s should be clearly minimized to easily manage the installation, so that it is quite natural to select for s the minimum value, able to guarantee the TL objective. The performance of the various configuration can be easily compared, from the curves displaying versus the hop length L: - the minimum PPTX power required to meet the POUT objective, and - the corresponding minimum Φo values, required to meet the TL objective

FIGURES 10 and 11 namely provide these results for the examined configurations A) through

D).

Besides doubling the average spectral efficiency, the advantage offered by the MIMO configurations over the conventional one is evident in terms of required minimum PPTX at equal hop length L.

As expected, the "Ix" modality with the lxNS(2,2) combined (TX+RX) diversity outperforms the lxNS(l,2) alternative of simple RX diversity. Even option D) with PHYmin = 16QAM turns out to be very interesting, because allowing to double the PHYmin throughput (^ 4xEl), while requiring about the same PPTX as the conventional STND(1,1) solution. A detailed comparison is given in TABLE 5. TABLE 5 Performance comparison at equal PPTX (= 15dBm)

Antenna separations are not high at all as shown by FIGURE 12 for configuration A).

FIGURE 12 shows two sets of curves: one in the assumption of adopting a "Constant Φo" design for all hop lengths below Lmax, the other one by adopting a "Minimum Φo" design, yielding the minimum separation able to meet the TL objective.

For the more interesting B) configuration, one obtains equal values at equal hop length for the minimum separations, while this option allows of course to reach a longer maximum distance

Lmax. For PPTX values in the range from 10 to 20 dBm, the separation is less than 2,3m and can even be significantly smaller for hop lengths lower than Lmax.

Exemplary embodiment : In the following the description is provided of an exemplary, possible embodiment of the system, mainly with the purpose of highlighting the technical problems and to discuss the relevant technical solutions. The described embodiment has to be considered "typical" in the sense that, even if alternative solutions would be available or possible, they would have anyhow to guarantee the practical realization of the process with the same functionalities and characteristics as commented here below

System operations are described for a single transmission direction (the GO direction), the other one (the REVERSE direction) being exactly equivalent.

The transmit signal is represented by the combination of the useful information signal (the data

Payload) and of the Service/Telemetry channels (having a rate obviously much lower than the payload), to be utilized for the transmission of one (or more) end-to-end Service Channels (EOW

= Engineering Order Wire) and of controls and commands (the "Telemetry") from one side to the other of the link.

Such combination of signals is usually obtained by means of well known digital multiplexing techniques, by realizing the segmentation of the "on air" signal in successive frames, all of the same duration and structure.

The segmentation is such that a small portion of the frame (the "service field") is reserved first to the required UWs (U and A), then to the error correction (FEC), and finally to the other services

(SRV). The remaining and predominant portion of the frame is reserved to the data Payload.

Clearly the quote left to all the services represents an overhead (OH) with respect to the payload

(PD), and both define the "Frame Efficiency" in the time domain, given by :

Frame Efficiency (%) = 100*PD/(PD+OH)

The detection of the unique word U allows for the frame synchronization of the receiver. The processing required for the creation of the "on air" framing and for the insertion of the services ("bit insertion") is performed by Block 201, TX PHY Level Processor, which also includes the functionalities of FEC Encoding, Dual Modulation and TX Distribution to the two antennas for the generation of the PHY Level in the transmit side. Block 201 generates two modulated signals for Block 202, Dual RF transmitter. The real time selection of the PHY Level is performed under the control of Block 203, Telemetry

Channel Receiver and Agent for the Link Adaptivity in the transmit side.

The two signals are then translated to the wanted RF frequency through IF/RF conversions and suitably amplified by Block 202, that includes the two RF transmitters feeding the relevant antennas. At the receive side the signals received by the two RX antennas are first amplified by low-noise amplifiers and then re-translated down at intermediate frequency through RF/IF conversions in

Block 204, Dual RF Receiver. Subsequently the signals are processed by Block 205, RX Combiner/Canceler and RX PHY Level Processor (Dual Demodulation, FEC Decoding and Telemetry extraction).

It is evident how, for a correct operation of the link, be it necessary to dynamically ensure the fine and accurate real time precision of the coordination between PHY Level selection at the transmit side and PHY Level assignment at the receive side. This feature of " TX/RX PHY Level Synchronization" is realized by keeping the two PHY Level Processors 201 and 205 in mutual tracking, via the remote telemetry channel.

The propagation conditions over the hop are estimated in real time through the measure of the received signal quality, derived from an indicator monotonically linked to the instantaneous bit error rate BER (CSI = Channel State Indicator).

Methods to obtain the real time estimate of the CSI include the measure of one of the following parameters:

- Receive Signal-to-Noise Ratio (S/N), simply derived from the measure of the signal power S received on the two antennas, being the noise N known and anyhow a constant for every channel bandwidth option (B W)

- Average number of corrections performed by the FEC decoding process counted in a convenient time interval (if a FEC coder/decoder is available as part of the PHY Level).

- MSE (Mean Square Error) for both demodulators, connected on the first and on the second antenna, respectively, measured at the decision points, after that individual, fine AGC stages have brought the receive signals Sl and S2 at a suitable and equal reference level for both the receive chains.

The real time CSI estimate is performed in Block 206, Channel Quality Measure and provided to

Block 207.

This last performs the functions reserved to the Controller (Manager) of the Link Adaptivity, acts also as Agent for the Adaptivity at the receive side and as Transmitter for the Telemetry

Channel.

Block 207 has the intelligence to manage the link adaptivity by controlling : - from remote, the distant transmission side (Block 201, through Block 203) via the the telemetry channel (Block 208, Return Telemetry), and

- locally, the near receive PHY Level Processor (Block 205).

To Block 207 external parameters are also provided (normally SW configurable parameters), representing programmable thresholds to be compared with the real time value of the estimated

CSI in order to dynamically derive a real time criterion for the selection of the PHY Level.

A precise time synchronism is required between the PHY Level switchover in the transmit side

(Block 201) and the corresponding update of the configuration at the receive side (Block 205), in order to avoid error bursts at every switchover. A suitable (but not critical) communication protocol is needed on the Telemetry Channel (MAC

= Medium Access Control), to ensure such precise and reliable TX/RX PHY Level synchronization.

The TX/RX PHY Level synchronization can for instance be implemented by locating exactly at the beginning of each frame the instants for possible switchovers, by cyclically numbering the successive frames and by transmitting to the processor 201 the frame number selected for the next switchover. For better reliability a backward confirmation message can be foreseen, before actuating the switchover.

For better clarity FIGURE 14 presents a typical PHY Level adaptation process, assuming a 2- state profile [PHYmin = 1x4(2,2) and PHYmax = 2x64(2,2)] and an exemplary hop with KK = 5OdB.

The characteristics of the profile are again recalled by TABLE 6

TABLE 6 : 2-state profile on an exemplary hop (KK=50dB)

FIGURE 14 shows in particular the behaviour of the receive (S/N) ratio as a function of the supplementary attenuation As over the hop.

By the increase of As with the increasing rainfall intensity, the available (S/N) ratio at the receiver proportionally decreases in dB. This ratio is monitored with adequate precision through the CSI estimate performed by block 206 (FIGURE 13). As soon as the CSI value drops below a certain threshold s2, the controller 207 decides to switch the PHY Level down the more robust

PHYmin.

The threshold s2 is given by : s2 = β₂(avg) + m (dB) where : β₂(avg) is the threshold (S/N)avg of the PHYmax m is a proper safety margin.

Controller 207 sends to the remote terminal 201 the first available frame number for the switchover and configures the local receiver 205 to switch at the same time.

This way the switchover occurs when As exceeds the value C (about 2OdB, in the example), the instantaneous throughput becomes drastically lower, but the link threshold also goes down and is reached only when the As value increases up to a value (about 47dB in the example) for which the available (S/N)avg becomes equal to βi(avg), the PHYmin threshold. Since the system is operated at constant static peak power PPTX on RF transmitters 202, the

(S/N)avg value goes instantaneously up in the change (PHYmax ^ PHYmin) by the "static" ratio ΔPTX between the average powers of the two modes (in our case ΔPTX = 3,68dB, ratio between 4QAM and 64QAM).

In the opposite sense, that is when As lowers with the hop recovering better propagation conditions, it is convenient to locate the new switching point back to PHYmax in correspondance of a different and higher (S/N)avg value, creating this way a suitable hysteresis zone for the switchovers in the two directions (increasing or decreasing As).

This solution allows to avoid oscillations between the two modes, oscillations that could occur for "quasi- stationary" or slowly changing propagation conditions.

To be noted also that the PHY Level Adaptivity can be combined to the automatic control of the transmit power (ATPC = Automatic Transmit Power Control), as shown by the same FIGURE

14.

The ATPC is a well known technique allowing to dynamically reduce the transmit power PPTX when As is particularly low in good propagation conditions (situation by the way occurring for a very high percentage of the time). The benefit is to reduce the interference level generated on other links utilizing the same RF frequency and located in the same geographical area.

For this feature it is possible to realize a "joint" adaptivity management (that is with ATPC combined to PHY Level adaptivity), with controller 207 in charge of such integrated management and with a proper MAC protocol to handle the dual functionality over the telemetry channel 208. With this kind of "joint adaptivity" five different zones of automatic operations can be distinguished for the link, as shown by FIGURE 14.

1) ATPC Zone : for low As values (from 0 to point A) PPTX grows proportionally to As. The PHY Level is at PHYmax and the PPTX value is controlled in remote loop via the MAC channel so as to maintain the (S/N)avg value constant and equal to a proper value β* = [β2(avg) + m*] , higher than β2(avg) by a convenient margin m*.

2) Zone at PHYmax : for values of As in the range from point A to point B. In this zone the link still operates at PHYmax with a constant PPTX value, equal to the maximum available transmit power.

3) Hysteresis zone : for values of As in the range from point B to point C. In this zone the switchovers between PHYmax and PHYmin occur for conveniently different (S/N)avg values, depending on the sense of the As variation, increasing or decreasing.

4) Zone at PHYmin : for values of As in the range from point C to point D. In this zone the link operates at PHYmin with a PPTX value equal to the maximum available one. As can continue to grow until the link threshold (point D) is exceeded and the link enters the next zone 5).

5) Outage zone : beyond point D, that is for (S/N)avg becoming lower than the PHYmin threshold βl(avg). Note that in the example of FIGURE 14 this happens for As exceeding about 47dB, thus with a very low probability.

Polarization reuse : So far, the proposed solution has been described with reference to a Single

Polarization configuration, using the best one available, that is the Vertical one.

It is however very possible to utilize both polarizations, with a second system added on the

Horizontal Polarization, this way further duplicating the spectral efficiency. This option would allow to realize an overall by-4 throughput multiplication, with respect to the conventional solution STND(1, 1) on single polarization, as likely used by most of the next generation products.

The number of dimensions (ND) on which the system operates becomes ND = 4, instead of the single polarization case of ND = 2. As for the receive processing, whose task is always to orthogonalize all the available dimensions, a conceptually simple "dimensionality expansion" has to be applied, in the sense of cancelling for each signal output the (ND-I) existing interference components.

For ND = 2, a single copolar interference component arises from the other antenna, and one single stage of copolar cancellation is required (CPIC = Copolar Interference Canceler), as described so far.

For ND = 4, three interference components arise, of which one is again copolar from the other antenna, and two are crosspolar from the same and from the other antenna. Three stages of cancellation are consequently required for each signal output, one CPIC and two of the XPIC type (XPIC = Crosspolar Interference Canceler). To cope with the higher number of interferers, also a higher precision is required for the signal quantization and for the signal processing arithmetic.

As extensively described in this Application, the Point-to-Point system and the relevant operating method according to the present invention allow to satisfy challenging requirements and to overcome the problems and limitations commented in the initial part of this description, relevant to the state-of-the art technology.

To satisfy specific or particular requirements, an expert of the art would possibly be able to introduce modifications or variants to the system embodiment here proposed and described, but it is believed that all such modification or variants will fall within the invention protection frame as defined in the following claims.

Claims

1. Point-to-point radio system comprising a plurality of transmit antennas and a plurality of receive antennas, said antennas being utilized with a transmission modality selected among a plurality of available modalities and being intended to transmit and receive single-carrier modulated signals on RF frequencies above 6GHz, said signals having a modulation/coding format selected among a plurality of available modulation/coding formats, said transmission modality of the antennas and said modulation/coding format defining jointly a transmission physical level, said system being characterized by the fact that is comprises, both in the transmit and in the receive sides, signal processing means respectively coupled to the inputs of the transmit antennas and to the outputs of the receive antennas and able to switch the transmission physical level as a function of the receive quality of the signal received by said plurality of receive antennas.

2. Point-to-point radio system according to Claim 1, wherein said signal processing means select and switch the physical level among two or more transmission formats, as a consequence of the variation of quality of the received signal, while maintaining the occupied RF bandwidth unchanged.

3. Point-to-point radio system according to Claim 1 or Claim 2, wherein said plurality of available transmission modalities includes the modality of spatial multiplexing and at least one between the modality of simple receive diversity, in which one transmit antenna is off, and the modality of combined transmit and receive diversity.

4. Point-to-point radio system according to any of the claims from Claim 1 to Claim 3, wherein said plurality of modulation/coding formats include different modulation formats with same FEC coding/decoding scheme, same modulation formats with different FEC coding/decoding schemes and different modulation formats combined to different FEC coding/decoding schemes

5. Point-to-point radio system according to any of the claims from Claim 1 to Claim 4, which operates in propagation environments predominantly causing non-selective (flat) attenuation versus frequency within the RF channel band.

6. Point-to-point radio system according to Claim 5, operating over high frequency RF links wherein said propagation environments cause rain attenuation or multipath attenuation with echo delays much smaller than the inverse of the channel bandwidth.

7. Point-to-point radio system according to any of the claims from Claim 1 to Claim 6, wherein said signal processing means switch the physical level by making use of the estimate of the channel quality through a "real time" measurement of a receive parameter monotonically linked to the instantaneous bit error rate of the signals received by the receive antennas.

8. Point-to-point radio system according to any of the claims from Claim 1 to Claim 7, wherein the estimate of the channel quality is realized by measuring at least one variable selected among

- the receive Signal-to-Noise Ratio (S/N), simply derived from the measure of the signal power S received on each of the receive antennas,

- the average number of corrections performed by the FEC decoding process, counted over a convenient time interval,

- the Mean Square Error (MSE) at the decision point for each of the demodulators connected to the receive antennas.

9. Point-to-point radio system according to any of the claims from Claim 1 to Claim 8, wherein the switchover between the plurality of transmission modalities is realized using a method based on at least two or more independent Unique Words (UW) inserted in the digital signal frames transmitted on the two antennas.

10. Point-to-point radio systems according to Claim9, wherein the use of at least two or more independent Unique Words is finalized to the purposes of maintaining the continuity over time of the feedback loops involved in the control of all the variables required for the correct operations of the various available transmission modalities, and of realizing the mutual independence of such controls and their complete decoupling from the physical level switchovers.

11. Point-to-point radio system according to any of the claims from Claim 1 to Claim 10, wherein said antennas can also be utilized on orthogonal polarizations.

12. Point-to-point radio system according to all or part of the Claims from 1 to 11, whereby said system is used on one polarization together with a second and equal system operated on an orthogonal polarization, with the final scope of multiplying by four the overall transmission capacity.

13. Method to operate a point-to-point radio system comprising multiple transmit and receive antennas, said antennas being utilized with a transmission modality selected among a plurality of available modalities and being destined to transmit and receive single-carrier modulated signals on RF frequencies above 6GHz, said signals having a modulation/coding format selected among a plurality of available formats, said transmission modality and said modulation/coding format defining jointly a transmission physical level, said method comprising switching, in transmit and in receive sides, the transmission physical level as a function of the receive quality of the signal received by said plurality of receive antennas.

14. Method according to Claim 13, wherein the physical level is switched among two or more transmission formats, as a consequence of the change of quality of the received signals, while maintaining the occupied RF bandwidth unchanged.

15. Method according to Claim 13 or Claim 14, wherein said plurality of available transmission modalities includes the modality of spatial multiplexing and at least one between the modality of simple receive diversity, in which one transmit antenna is off, and the modality of combined transmit and receive diversity.

16. Method according to any of the claims from Claim 13 to Claim 15, wherein said plurality of modulation/coding formats include different modulation formats with same FEC coding/decoding scheme, same modulation formats with different FEC coding/decoding schemes and different modulation formats combined to different FEC coding/decoding schemes

17. Method according to any of the claims from Claim 13 to Claimlβ, for radio systems which operate in propagation environments predominantly causing non-selective (flat) attenuation versus frequency within the RF channel band.

18. Method according to Claiml7, for radio systems operating over high frequency RF links wherein said propagation environments cause rain attenuation or multipath induced attenuation with echo delays much smaller than the inverse of the channel bandwidth.

19. Method according to any of the claims from Claiml3 to Claiml8, wherein the physical levels are switched by making use of the estimate of the channel quality through a "real time" measurement of a receive parameter monotonically linked to the instantaneous bit error rate of the signals received by the receive antennas.

20. Method according to any of the claims from Claim 13 to Claim 19, wherein the estimate of the channel quality is realized by measuring at least one variable selected among:

- the Receive Signal-to-Noise Ratio (S/N), simply derived from the measure of the signal power S received on each of the receive antennas, - the average number of corrections performed by the FEC decoding process counted over a convenient time interval,

- the Mean Square Error (MSE) at the decision point for each of the demodulators connected to the receive antennas.

21. Method according to any of the claims from Claim 13 to Claim 19, wherein the switchover between the plurality of transmission Modalities is realized using a method based on at least two or more independent Unique Words (UW) inserted in the digital signal frames transmitted on the two antennas.

22. Method according to Claim 21, wherein the use of at least two or more independent Unique Words is finalized to the purposes of maintaining the continuity over time of the feedback loops involved in the control of all the variables required for the correct operations of the various available Transmission modalities, and of realizing the mutual independence of such controls and their complete decoupling from the Physical Level switchovers.

23. Method according to any of the claims from Claim 13 to Claim 22, for radio systems wherein said antennas can also be utilized on orthogonal polarizations.

24. Method according to any of the claims from Claim 13 to Claim 23, for a radio system used on one polarization together with a second and equal system operated on an orthogonal polarization, with the final scope of multiplying by four the overall transmission capacity.