Rate-splitting multiple access (RSMA)

Abstract

A transmitter encodes incoming messages from users; splits each encoded message into a common (combinable) part and a private (not combined) part having a number of elements (e.g., bits or symbols) ba

Description

FIELD OF THE INVENTION This disclosure describes a system and method for Rate-Splitting Multiple Access (RSMA), and more particularly Codeword-Segmentation RSMA (CS-RSMA). The invention relates to a transmitter, a receiver, a communication system comprising the transmitter and the receiver, a transmission method, a receiving method, and a communication method comprising the transmission method and the communication method. BACKGROUND TO THE INVENTION In the context of mobile telecommunications, High Demand Density (HDD) scenarios occur in situations where a lot of users and devices are present in a specific place or event, eg dense urban areas, airports, sports venues, railway and subway stations and major public events. Around 1% of UK geographical areas served by mobile networks could be characterised as High Demand Density (HDD) areas, which drive an estimated 20% of total mobile traffic. However, mobile telecommunication systems face particular challenges in such areas, due to the intense demand on the network. To increase network capacity, Multi-user multiple-input and multiple-output (MU-MIMO) systems generally use the spatial domain, with multiple antennas at transmitters and receivers, to multiplex users in space. It is widely implemented in wireless networks, for example, 5G cellular networks. Indeed, MU-MIMO is generally a key technology for 5G NR (New Radio), for supporting large number of users with high data rates. MIMO enhancements for NR have been considered extensively at 3GPP TSG RAN since Rel-15. However, challenges remain for MU-MIMO in High Density Demand areas and / or regarding inaccurate CSIT (Channel State Information at the Transmitter). For example, when using MU-MIMO with 5G NR, close proximity of lots of users can result in a situation wherein channels are not orthogonal to each other and interference arises. Rate-Splitting Multiple Access (RSMA) is an interference management technology that enhances the spectrum efficiency of MU-MIMO wireless networks. Currently, RSMA systems are being studied extensively in academia and being considered by telecommunication industry, e.g., for 5G or upcoming 6G systems. The performance benefit of RSMA over standard approaches included in 5G has been demonstrated with numerical simulations. Generally, an RSMA system splits incoming messages into common and private parts before encoding into codewords / data streams, the encoding comprising adding redundancy. (Merely for assistance, it is noted that the term ‘common’ as used herein generally means ‘combined’ or ‘combinable’, whereas the term ‘private’ generally relates to data that is not to be combined with other data. Thus, for example, the terms do not imply any presence or absence of data privacy as such. On the other hand, ‘common’ and ‘private’ generally refer to message parts, eg codewords or data streams, that are intended for (to be received by), respectively, multiple users or a single user). Fig. 1 illustrates a transmitter of such a system, wherein messages I / I / 1 - Wk are intended for (or requested by) respective users and split into private parts Wp and common parts Following combining of common parts into a combined message by a message combiner, an encoder receives the combined message and the uncombined private parts, thus receiving A"+l messages. The received K+l messages are then encoded by the encoder, and passed through an output stage comprising modulation and precoding for onward transmission. (Though encoding and modulation may in some implementations be done together, i.e., such that one combined system block I circuit generates symbols from uncoded bits). For more detail, Figs. 2A - 2C illustrate PHY layer configurations of such a system, wherein Fig. 2A shows a transmitter and Figs. 2B and 20 show alternative receiver configurations. Specifically, each block immediately preceding a multiplexer (i.e., combiner) of Figs. 2B and 20 is a hard-decision quantizer, where the LLRs which are real numbers {-^,+°0} that are quantized to 0 if LLR<0 or are quantized to 1 if LLR >1 (or the other way around depending on LLR definition). We further note that the forked lines in Fig. 2B may be considered as two output lines from MA-1. In both Fig. 2B and 20 there is no splitter: presence of a splitter may depend on how the computation works; they can be naturally separated. More specifically, at the transmitter side of such a system: 1) The transmitter first splits each of the K unicast messages, into common and private parts, i.e., Wi is split into W(c,V) and fF(p,l), ^2 is split into jy(c,2) and W<p,2)______ % is split into fT(c,A) and W(p,K). 2) All the common parts, {lV(c,l),...,l4 / (c, / Q}, are combined into a common message, We, while the private parts, {W(pX),...,W(p,Ky}, remain as separate messages. 3) The resulting K+l messages, Wc and are encoded into K+1 data streams followed by modulation, precoding and transmission. Notwithstanding the above, there remains a need to, inter alia, improve performance (e.g., increase throughput, reduce latency, improve interference management), reduce complexity and / or increase efficiency (e.g., signalling overhead) of an RSMA communications system and / or method. For use in understanding the present invention, the following disclosures are referred to: C1. “Rate-Splitting Multiple Access: Fundamentals, Survey, and Future Research Trends”, Mao et al, 30.9.2022, available at https: / / arxiv.org / abs / 2201.03192; C2.: “A Primer on Rate-Splitting Multiple Access: Tutorial, Myths, and Frequently Asked Questions", Clerckx et al, 10.1.23, available at https: / / arxiv.org / abs / 2209.00491; and C3.: “Rate-Splitting Multiple Access: Finite Constellations, Receiver Design, and SIC-free Implementation”, Zhang et al, 7.12.23, available at https: / / arxiv.org / pdf / 2305.17178. SUMMARY According to a first aspect of the present invention, there is provided a transmitter comprising: an encoder to encode incoming messages from respective said users of the transmitter; a stream splitter to split each encoded message to form from each split message a common part and a private part, each of the parts having a number of elements such as bits or symbols, wherein each said splitting of a said encoded message is based on a splitting order that determines elements of the encoded message that are to form the common part of that encoded message and elements of the encoded message that are to form the private part of that encoded message, wherein the number of elements of each of the common part and the private part of at least one of the split encoded messages is more than or equal to one; a stream combiner to form a common stream comprising each said formed common part of the encoded messages, wherein the forming is based on a combining order that determines where each element of each common part is positioned in the common stream; and an output stage for outputting signals for transmission based on the common stream and private parts. The encoding generally comprises adding redundancy. Advantageously in this regard, embodiments of the transmitter, which is generally for RSMA, may thus first encode independent messages into encoded data streams and then split those data streams into common and private parts. This may allow a corresponding receiver to demodulate only a segment of the common stream, combine it with the desired private stream symbols and decode the combined stream. In an embodiment, the splitter may be configured to perform the splitting of at least one of the encoded messages such that the number of elements of the common part of the encoded message is zero, and / or to perform the splitting of at least one of the encoded messages such that the number of elements of the private part of the message is zero. However, in embodiments, all incoming messages from respective users may be split. In this case, the transmitter comprises: an encoder to encode incoming messages from respective said users of the transmitter; a stream splitter to split each encoded message to form a common part and a private part each having one or more elements such as bits of symbols, wherein the splitting is based on a splitting order that determines which one or more elements of an encoded message form the common part; a stream combiner to combine the common parts of the encoded messages to form a common stream, wherein the combining is based on a combining order that determines where each element of each common part is positioned in the common stream; and an output stage for outputting signals for transmission based on the common stream and private parts. In any embodiment, the same splitting pattern may be applied for all transmitter users, or different splitting patterns may be applied for at least one of respective users. These patterns(s) may be fixed within a system or communicated eg via signalling to respective receivers, so that a matching desplitting order can be applied at each receiver. Generally, the transmitter may further comprise modulation, precoding and / or resource mapping blocks. It is further noted that any above-mentioned symbols may for example be constellation symbols, for example QAM, eg QPSK, symbols. There may further be provided the transmitter for unicast communication, wherein the incoming messages are unicast messages. There may further be provided the transmitter for unicast and non-unicast communication, wherein the incoming messages comprise unicast messages and the transmitter is further for transmitting a non-unicast message, wherein non-unicast is multicast or broadcast, and wherein: the encoder is to encode the unicast and nonunicast messages; and the stream combiner is to combine the encoded non-unicast message and each said formed common part of the encoded unicast messages to form the common stream, wherein the combining is based on a combining order that determines where each element of each common part and of the encoded nonunicast message is positioned in the common stream. (References herein to ‘non-unicast’ generally mean multicast and / or broadcast. Furthermore, references herein to ‘multicast’ and ‘broadcast’ are generally interchangeable. Thus, an embodiment described as being for multicast may alternatively or additionally be for broadcast. Preferred embodiments provide unicast and multicast communication, or unicast and broadcast communication). There may further be provided the transmitter, comprising: a modulation stage to modulate the common stream and the private parts and forward the modulated stream and parts to the output stage, wherein the outputting is based on the forwarded stream and parts, preferably wherein the output stage comprises a precoder to precode signals based on the modulated stream and parts and to form the outputted signals based on the precoded signals. We further describe the transmitter, optionally comprising: a modulation stage to modulate the encoded messages and forward each modulated message to the stream splitter, wherein the splitting splits the forwarded messages to form the common and private parts, preferably wherein the output stage comprises a precoder to precode signals based on the formed stream and parts, wherein the outputted signals for transmission are based on the precoded signals. There may be provided the transmitter comprising: a modulation stage to modulate encoded unicast messages from the encoder and forward each modulated unicast message to the stream splitter, wherein the splitting splits the forwarded messages to form the common and private parts, preferably wherein the transmitter is for unicast and non-unicast communication as described above and the modulation stage is further to modulate an encoded non-unicast messsage from the encoder and forward the modulated non-unicast message to the stream combiner, preferably wherein the output stage comprises a precoder to precode signals based on the formed stream and parts, wherein the outputted signals for transmission are based on the precoded signals. There may further be provided the transmitter to, in response to receiving a message retransmission request, retransmit a said common stream and private part having elements of a message corresponding to the message retransmission request, preferably wherein the request and response are in accordance with a HARQ process. According to a second aspect of the present invention, there is provided a receiver comprising: an estimator to generate modulation symbol detection results of a received common signal and to output the results as an estimated common stream, and to generate modulation symbol detection results of a received private signal and to output the results as an estimated private part; a stream decomposer to extract from the estimated common stream results to form an estimated common part corresponding to the receiver user, wherein the extracting is based on a decomposing order; and a stream combiner to combine the estimated common part corresponding to the receiver user and the estimated private part to generate an estimated message stream, wherein the combining is based on a desplitting order. The generated message stream may decoded subsequently. The receiver may be of various types, for example a joint de-mapping receiver or a soft Symbol Level Interference Cancellation, SLIC, receiver. In embodiments, the estimation results can be estimates of transmitted symbols or may take a different form. For example, the results can be calculated as loglikelihood-ratios (LLRs); such a ratio may be interpreted as a soft result from symbol detection, rather than the symbol itself. More broadly, any form of results that carries knowledge acquired from estimation can be used. Thus, the estimator may generate estimated symbols as such, or for example LLRs. In this regard, and in a specific example, the estimator may be to estimate modulation symbols of the received common signal and to output the estimated symbols as the estimated common stream, and to estimate modulation symbols of a received private signal and to output the estimated symbols as the estimated private part. The stream decomposer may extract symbols from the estimated common stream symbols to form the estimated common part corresponding to the receiver user, wherein the extracting is based on the decomposing order. The stream combiner may combine the estimated common part and the estimated private part to generate the estimated message stream, wherein the combining is based on the desplitting order, wherein the receiver can be implemented in various ways, such as a joint de-mapping receiver or a soft Symbol Level Interference Cancellation, SLIC, receiver. In an embodiment with a joint de-mapping receiver, the receiver may be configured to first jointly de-map the received signals into coded-bit-LLRs of common and private streams. The LLRs may be multiplexed to form the LLRs of a requested codeword and fed into decoders to obtain uncoded bits of the requested message. In an alternative embodiment using soft SLIC (Symbol Level Interference Cancellation) however, the coded-bit-LLR for the common stream may first be computed by a demapper (de-modulator). These LLRs may then be used to compute expectation of the common stream symbols, also called soft symbols. Next, the receiver may subtract the soft symbols from the received signals and de-map (or demodulate) the private streams. Subsequently, the LLRs may be multiplexed to form the LLRs of the requested codeword and fed into decoders to obtain uncoded bits of the requested message. Advantageously, embodiments of the receiver, which are generally for RSMA, may demodulate the entire common stream and extract the desired segment from the demodulated common stream but it is not required to decode it. The desired demodulated common stream segment may be combined with the desired private stream symbols, and the combined stream decoded into the requested message. Generally, the receiver may further comprise precoding and / or resource demapping blocks. It is further noted that any above-mentioned symbols may for example be constellation symbols, for example QAM, eg QPSK, symbols. There may further be provided the receiver for unicast communication, wherein the estimated message stream comprises an estimated unicast message. There may further be provided the receiver for unicast and non-unicast communication, wherein non-unicast is multicast or broadcast, wherein: the extraction by the stream decomposer further extracts from the estimated common stream results to form an estimated non-unicast stream, wherein the extraction to form the estimated common part and the estimated non-unicast stream is based on the decomposing order. There may further be provided the receiver, comprising a decoder to decode the generated estimated message stream and to output a message based on the decoded stream, preferably wherein the decoder performs FEC decoding such as LDPC decoding. There may further be provided the receiver, wherein the decoder is further to decode the estimated non-unicast stream and to output a non-unicast message based on the decoded estimated non-unicast stream. There may further be provided the receiver, wherein the decoder is to detect an error in estimated stream corresponding to a message, and wherein the receiver is to, in response to a said error detection, transmit a request for retransmission of a common stream and private part corresponding to the message, the retransmission request preferably in accordance with a HARQ process. A communication system may be provided comprising the transmitter and at least one receiver, the transmitter and receiver(s) as defined above. Thus, an embodiment may comprise an end-to-end system for RSMA. There may further be provided the communication system, wherein the decomposing order of the receiver matches the combining order of the transmitter. There may further be provided the communication system, configured to communicate to the receiver an indication of the combining order of the transmitter. There may further be provided the communication system, wherein the desplitting order of the receiver matches the splitting order of the transmitter. There may further be provided the communication system, configured to communicate to the receiver an indication of the splitting order of the transmitter. There may further be provided the communication system, wherein the transmitter and receiver are to perform a HARQ process. There may further be provided the communication system, wherein the system is a MU-MIMO system, such as a massive MIMO system, preferably wherein the system is a multi-layer MIMO system such as for 4GLTE and 5GNR, configured to map the signal to be outputted for transmission based on the common stream and the signals to be outputted for transmission based on the private parts to multiple MIMO layers. There may further be provided the communication system, for use in a cellular telecommunications, wifi or satellite communications network, such as a 5G NR network. There may further be provided the above-mentioned transmitter, receiver or communications system, for RSMA such as cooperative rate splitting RSMA, hierarchical rate splitting RSMA or multilayer splitting RSMA. According to a third aspect of the present invention, there is provided a transmission method comprising: encoding incoming messages from respective users into codewords; and splitting at least one of the encoded codewords to form from each split codeword a common part and a private part, wherein each said splitting of a codeword is performed according to a splitting order, wherein the common part and the private part formed from at least one encoded codeword each comprise one or more elements of the codeword; forming a common stream comprising each split common part of the at least one of the encoded codewords, wherein the forming comprises combining elements of the common parts according to a combining order; outputting signals for transmission based on the common stream and private parts. In embodiments, encoded messages from all users may be split such that neither of the common part and private part of each codeword is, in effect, empty. In other words, one or more elements of each encoded code word are put into the private part and at least one remaining element into the common part. In this case, and more generally, the transmission method comprises: encoding incoming messages from respective users into codewords; and splitting each encoded codeword into a common part and a private part, wherein the splitting is performed according to a splitting order; combining the split common parts of the encoded codewords into a common stream, wherein the combining is performed according to a combining order; outputting signals for transmission based on the common stream and private parts. Regardless, embodiments of the transmission method are generally for RSMA. There may further be provided the transmission method, for unicast and non-unicast communication, wherein the incoming messages comprise at least one unicast message and a non-unicast message, wherein non-unicast is multicast or broadcast, wherein: the encoding encodes the unicast and non-unicast messages; and the combining combines the common parts of the encoded unicast messages and the encoded non-unicast message to form the common stream, wherein the combining is based on a combining order that determines where each element, of each common part and of the encoded non-unicast message is positioned in the common stream, preferably wherein the elements are either bits or symbols. According to a fourth aspect of the present invention, there is provided a receiving method comprising: receiving a common signal and a private signal; generating modulation symbol detection results of the common signal and outputting an estimated common stream based on the results; generating modulation symbol detection results of the private signal and outputting an estimated private part based on the results; extracting from the estimated common stream a common part for a respective user, the extraction based on a decomposing order; combining the estimated private part and the extracted common part to generate a message stream, the combining based on a desplitting order; and then decoding the generated message stream. Generally, embodiments of the receiving method are for RSMA. In example embodiments, the method may be performed by a joint de-mapping receiver or a soft Symbol Level Interference Cancellation, SLIC, receiver. Similarly as for the receiver discussed above, the estimation results can be estimates of transmitted symbols or may take a different form. Thus, any detection results may be estimated symbols as such, or for example LLRs. In this regard, and in a specific example, the receiving method may comprise: receiving by a receiver a common signal and a private signal; estimating modulation symbols of the common signal to output an estimated common stream, and estimating modulation symbols of the private signal to output an estimated private part; extracting from the estimated common stream a common part for a respective user, the extraction based on a decomposing order; combining the estimated private part and the extracted common part to generate a message stream, the combining based on a desplitting order; and then decoding the generated message stream. There may further be provided the receiving method, for unicast and non-unicast communication, wherein non-unicast is multicast or broadcast, wherein: the extracting further extracts from the estimated common stream symbols to form an estimated non-unicast stream, wherein the extraction to form the estimated common part and the estimated non-unicast stream is based on the decomposing order; and the decoding decode the generated message stream to output a message, and decodes the estimated non-unicast stream to output a non-unicast message. A communication method may be provided, comprising any above-mentioned transmission method and any above-mentioned receiving method, wherein the signals transmitted based on the common stream and the transmitted signals based on the private parts are the received common and private signals, respectively. There may further be provided the communication method, comprising the transmission method and the receiving method, wherein the transmitter signals to the receiver a type of coding scheme used for the received common signal. The coding scheme may corresponding to a forward error correction (FEC) technique, eg LDPC. For example, in embodiment for unicast and non-unicast, the receiver may be informed of the coding scheme (e.g code-rate and / or code-word length of FEC such as LDPC) for the non-unicast part. In an embodiment for unicast only signals, the transmitter may inform the receiver about the modulation of the common stream. According to a related aspect of the invention, there is provided a data carrier (eg non-transitory) carrying processor control code which when running on a processor, causes the processor to implement any above-described method. Processor control code, to implement the above-described apparatus and methods, may for example be stored on a general purpose computer system or on a digital signal processor (DSP). The code may be provided on a carrier such as a disk, a microprocessor, programmed memory such as non-volatile memory (e.g. Flash) or read-only memory (Firmware), or on a transitory data carrier such as an optical or electrical signal carrier. Code (and / or data) to implement embodiments of the invention may comprise source, object or executable code in a conventional programming language (interpreted or compiled) such as C, assembly code, code for setting up or controlling an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array), or code for a hardware description language such as Verilog™ or VHDL (Very high speed integrated circuit Hardware Description Language). As the skilled person will appreciate such code and / or data may be distributed between a plurality of coupled components in communication with one another. An embodiment may comprise a controller which includes a microprocessor, working memory and program memory coupled to one or more of the components of the system. The or each processor mentioned above may be implemented in any known suitable hardware such as a microprocessor, a Digital Signal Processing (DSP) chip, an Application Specific Integrated Circuit (ASIC), Field Programmable Gate Arrays (FPGAs), etc. The or each processor may include one or more processing cores with each core configured to perform independently. The or each processor may have connectivity to a bus to execute instructions and process information stored in, for example, a memory. Preferred embodiments are defined in the appended dependent claims. Any one or more of the above aspects and / or any one or more of the above optional, i.e., preferable, features of the preferred embodiments may be combined, in any permutation. Furthermore, any of the above methods may be provided as corresponding apparatus, and vice versa. BRIEF DESCRIPTION OF THE DRAWINGS For a better understanding of the invention and to show how the same may be carried into effect, reference will now be made, by way of example, to the accompanying drawings, in which: Fig. 1 shows a transmitter of a prior art, i.e., known, RSMA system; Figs. 2A - 2C illustrate, at the physical layer (PHY), configurations of such a known system, wherein Fig. 2A shows a transmitter and Figs. 2B and 2C show alternative receiver configurations; Figs. 3A and 3B respectively show a transmitter and a receiver of a CS-RSMA system embodiment for multi-user unicast; and Figs. 4A and 4B respectively show PHY layer configurations of a transmitter and alternative receiver designs of a CS-RSMA embodiment for unicast-only; Figs. 5A and 5B respectively show another PHY layer design of a transmitter and alternative receiver configurations of a CS-RSMA embodiment for unicast; Figs. 6A and 6B respectively show a transmitter and a receiver of a CS-RSMA system embodiment for multi-user unicast and for multicast and / or broadcast, eg for non-orthogonal unicast and multicast and / or broadcast (NOUMB); Figs. 7A and 7B respectively show, at the PHY layer, a transmitter and alternative receiver configurations of a CS-RSMA system embodiment for multi-user unicast and for multicast and / or broadcast, eg for NOUMB; Figs. 8A and 8B respectively show, at the PHY layer, another design of a transmitter and alternative receiver configurations of an embodiment of CS-RSMA for multi-user unicast and for multicast and / or broadcast, eg NOUMB; Fig. 9 shows link-level unicast-only simulations of RSMA and of PHY transceiver embodiments of CS-RSMA, the simulations for systems using joint demapping and also systems using soft SLIC receiver types; Fig. 10 shows a general purpose computing device, on which the signal processing of a transmitter or receiver of an embodiment may be implemented. Within the drawings, features (eg paths and blocks) shown in orange and blue generally correspond to, respectively, common and private features. DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS Embodiments of the present invention are for a type of RSMA - the type referred to herein merely for convenience as Codeword-Segmentation RSMA (CS-RSMA) -wherein, generally speaking, a transmitter first encodes independent messages into encoded data streams and then splits those data streams into common and private parts. This contrasts with an RSMA system type wherein messages for transmission are split into common and private parts before encoding. In either type, coding may be implemented such that the encoding provides data streams comprising channel codes or codewords, such as may result from Forward Error Correction (FEC) eg LDPC encoding. Generally, CS-RSMA embodiments thus provide an RSMA architecture that may have similar performance to known RSMA implementations but may require less complexity at the transmitter and / or receiver, and / or allow communication with lower control signalling overhead. Such advantages, and / or others, may be achieved when applied for example in a MU-MIMO communications system (eg a single link or a network communicating transmitter(s) and receiver(s)), such as a network according to the 3GPP standard for 4G or 5G (and an expected 6G network), a IEEE 802.11 Wifi network or a satellite communications link. More generally, embodiments are widely applicable to communications networks including eg fixed, mobile, wireless (RF or free-space optical) and / or wired (electrical or fibre-optic). For example, embodiments may be implemented in applications identified in the above-mentioned document C2: further-enhanced mobile broadband (FeMBB), enhanced Ultra Reliable Low Latency Communication (eURLLC), or enhanced Massive Machine Type Communication (emMTC). Indeed, embodiments are generally applicable wherever a known RSMA could be alternatively used, e.g., Massive MIMO (eg FDD, TDD or Cell-free), Cellular IOT, Millimetre Wave / THz, Multigroup Multicast, unicast and multicast Mixed Services, Unmanned Aerial Vehicle, Satellite Communications, Multicell Networks, Physical Layer Security, Cooperative Transmission with User Relaying, Reconfigurable Intelligent Surface (RIS), Wireless Information and Power Transfer, Cloud / Fog RAN, Integrated Sensing and Communication (ISAC), Joint Communication and Jamming, Visible Light Communications (VLC), Low Latency and Finite Block Length, Cognitive Radio, Mobile Edge Computing, Federated Learning, and Network Slicing. We further note that RSMA can be viewed as a family of schemes that split messages into private and common parts. In this regard, CS-RSMA embodiments such as those described herein are generally applicable to replace any known RSMA variant, eg generalized rate splitting, cooperative rate splitting, hierarchical rate splitting or multilayer splitting. Generally in RSMA, the use of a common stream reduces the effects of interference. In particular, with some MU-MIMO systems, the coexistence of multiple data streams leads to multi-user interference, i.e., users receive undesired signals. With RSMA, the private streams is similar to the data streams in such MU-MIMO systems and suffer from multi-user interference, but the common stream is introduced. The common stream is intended for multiple users and hence do not create multi-user interference. Hence, by allocating part of the data payloads to the common stream, the reception of those data are not affected by multi-user interference. With flexibly allocating data and transmitter power to the common and private streams, RSMA generally achieves better performance (e.g. throughputs, latency) than non-RSMA MU-MIMO. Similar applies to CS-RSMA embodiments such as those described below, noting that: (a) with known RSMA, the receiver demodulates the entire common stream, decodes it and extract the desired part from the decoded message. This desired common part is to be combined with the decoded message from the desired private stream; and (b) in CS-RSMA embodiments, the receiver demodulates the entire common stream and extracts the desired segment from the demodulated common stream but it is not required to decode it. The desired demodulated common stream segment is to be combined with the desired private stream symbols and the combined stream is decoded. Figs. 3A and 3B show an embodiment of CS-RSMA for multi-user unicast, comprising a transmitter T1 and receiver R1-k, the constituent elements of each of which may be colocated or respectively located within a mobile device (eg phone, tablet or other user equipment) and / or base station (or vice versa) of eg a 5G network. More generally, the elements of the transmitter and / or of the receiver may be located at a single site or distributed across multiple sites, for example sites of a system wherein antennae are distributed over a wide geographical area. At the transmitter T1 of Fig. 3A, incoming messages M4 - WK are intended for (or requested by) respective K users and firstly encoded. The encoder ET1 may be for FEC, eg use polar codes or low-density parity-check (LDPC) encoding. The encoding may add systematic and / or parity bits. The encoded streams Xi - XK are then input to a stream splitter SST1 to output separate private parts Xp,i - XP,K and common parts Xi - k- k stream combiner SCT1 then combines the common parts (which may have the same or different numbers of bits for individual users 1 - K, for example depending on the number of bits in respective incoming messages) into a common stream X. Subsequently, the combined stream and private parts are input to a modulation stage M1 (for example for QAM eg QPSK modulation) before forwarding to an output stage preferably comprising the precoder PT1. The precoding preferably uses a linear precoder, and may reduce interference eg between streams or symbols. In embodiments, a precoding technique such as zero forcing may be applied to the private parts; a different technique may however be applied embodiments where the common stream is to reach and be decodable by all recipients. Other known MU-MIMO precoding techniques may be applied to each private part to allow decoding by the receiver corresponding to the intended recipient of the private part. For further detail, it is noted that linear equalization - as opposed to linear precoding - may be performed in any CS-RSMA transmitter embodiment, though for simplicity this is not shown in the drawings. It is assumed for embodiments described herein that the content and thus bits of the incoming messages are independent. In this regard, and while disclosure herein often focusses on the bit level - consistent with the above references to bits in relation to Fig. 3A - it is particularly noted that the stream splitting and stream combining operations can in any CS-RSMA embodiment be done at either bit level (before modulation) or at symbol level (after modulation). In CS-RSMA embodiments where stream splitting and stream combining operations are done at symbol level, the encoding and modulation stages can be done in a single stage, such that modulated symbols are calculated - in some cases directly - from incoming messages W4 - Wfc Stream combining and stream splitting (and corresponding receiver operations) may instead be performed at symbol level if the receiver knows the constellation order (e.g., whether QPSK, 16QAM) that has been used for all the combined streams. We additionally note that, in CS-RSMA embodiments where stream splitting and stream combining operations are done at symbol level, the encoding and modulation stages can be done in a single stage, such that modulated symbols are calculated from incoming messages WT - WK. Thus, embodiments may provide systems with joint modulation and coding (e.g., multilevel coding). It is further noted that, in any CS-RSMA transmitter embodiment the common stream Sc and / or private streams S(p,k) may be mapped to multiple MIMO layers (i.e., multilayer MIMO). Specifically, and considering eg a multi-layer MIMO system such as for 4GLTE and 5GNR, the transmitter may map a common stream Sc formed by the stream combiner of the transmitter and private parts Sp,1 - Sp,k formed by the stream splitter to multiple MIMO layers. Corresponding de-mapping may take place at the receiver. Consistent therewith, PHY layer drawings herein show resource mapping and de-mapping. Regarding in more detail the stream splitter SST1 of Fig. 3A, the selection of individual message bits as ‘common’ or ‘private’ is preferably independent of the content of the message. The splitting may be carried out simply based on an order of elements eg bits in each message. For example, a first part (such as the first m = eg 3 bits) or a first portion (eg half) of each message may be selected as a common part and the remainder as a private part, or a predetermined number of adjacent bit(s) of a message starting at every nth (eg n=3) bit may be selected as a common part and the remainder as a private part. Additionally or alternatively, the order - or splitting pattern - may select non-adjacent bits as common eg at least one every mth bit, or every mth bit and / or every nth bit etc. However, any other splitting order or pattern can be considered, eg selecting m bits randomly for the common stream, preferably wherein a matching desplitting operation is performed at the receiver, eg by the stream combiner SCR1 of the receiver R1-k as described below. The splitting pattern can also depend on alternative and / or additional criteria such as layer protocols, incoming message structure, etc.. (For assistance, we note that the term ‘order’ is generally used interchangeably herein relative to the term ‘sequence’ or ‘pattern’; thus an order may be a splitting, combining, desplitting or decomposing pattern as appropriate to the context). Similarly, and regarding the stream combiner SCT1 of Fig. 3A, the combining may use a combining order that concatenates (eg appends) respective users’ common parts in turn, interleaves eg successive x bit(s) of respective users’ common parts in turn, and / or applies logic operation(s) eg XOR to respective users’ common parts, to build the common stream. However, any combining order I pattern can be implemented provided that a matching decomposing operation is performed at each receiver. Further regarding the transmitter, and considering the signals xi xNT transmitted from respective antennae as shown in Fig. 3A, it is noted that, with the procedure of precoding / beamforming, each antenna may transmit a mixture of the K+1 streams. Hence, NT does not necessarily equal K+1. In embodiments, the K+1 streams (common and private), once transmitted, may not be separated in the frequency domain, but may instead be separated in the spatial domain (again, through precoding / beamforming), although different transmissions maybe separated in frequency and / or time domain. (Further noting that in some embodiments the common stream may not be separated, and may be superimposed to the private streams). For example, if each of the K+1 streams consists of N symbols: at the n-th transmission, the transmitter may transmit a symbol vector of length K+1, whose elements are the n-th symbols of the K+1 streams. This symbol vector may be precoded (eg comprising mixing at the transmitter antennas) and transmitted. Such a single transmission may happen within the same channel (eg same sub-carrier if OFDM is used, or same resource element with 3GPP’s standard). For N transmissions, an embodiment may use N channels (again, these can be created using OFDM and / or TDM). Hence, each receiver may look at all the N channels and decode its desired message from the mixture of symbols. Each receiver R1-k corresponding to a respective recipient / user (User-1 - User-K) as shown in Fig. 3B receives from the transmitter T1 the common stream and a private part for the recipient / user, wherein the common stream and / or private part may be received as one or more packets. These received signals are those that are receivable at the particular receiver dependent on the type of multiplexing, eg OFDM, implemented at the transmitter. It is noted in this regard that, in each channel, a received signal may be a mixture of common stream symbol and private stream symbols, in embodiments with change in amplitude and phase on each symbol due to effects of the precoding and / or the channel. With knowledge of the amplitude and phase change (eg based on a telecommunications standard), a receiver embodiment can discriminate the common stream symbol and the desired private stream symbol from a channel. Repeating this with all the channels of interest, such a receiver can estimate (eg by demapping and / or demodulation) the entire desired private stream and common stream segment. Decoding can be performed based on these estimates. Optionally, each receiver R1-k is a joint de-mapping receiver or a soft SLIC (symbol level interference cancellation) receiver. In this regard, an input stage of the receiver comprises an estimator ER1, generally for demodulation and / or demapping (see resource de-mapping and demodulation blocks M~1 in receiver PHY layer drawings herein). In an CS-RSMA system, the receiver will have a resource de-mapping block if resource mapping is done at the transmitter. The resource de-mapping is generally the first step at the receiver. The estimator ER1 may estimate modulation symbols (such as QAM, eg QPSK, symbols) of the received common stream and private part, and correspondingly output the an estimated common stream ScA and an estimated private part Sp,kA. A stream decomposer SDR1 of the receiver then extracts from the estimated common stream ScA the common part Sc,kA intended for the receiver user. (For convenience, corresponding notation is used throughout this disclosure: for example, sc is the same as ScA). Generally, the extraction is based on a decomposing order / pattern that matches the order in which basis the transmitter stream combiner SCT1 combined the common parts of respective (encoded) user messages to create the common stream Xc. That order, eg combining pattern, - or at least an indication thereof - may be signalled from the transmitter T1 to the receiver R1-k in a separate channel. Additionally or alternatively, the receiver may store, in embodiments permanently eg in ROM and / or fixed program code, that order (or indication) and / or the matching decomposing order. Regardless, a signalled and / or stored order is used to determine how the desired common part Sc,kA is separated from the common stream ScA for output by the stream decomposer SDR1. A stream combiner SCR1 then combines the receiver’s estimated and extracted common part Sc,kA (from the stream decomposer) with that receiver’s estimated private part Sp,kA (from the estimator), to output a message stream SkA to a decoder DR1. The combining may be based on a desplitting order or pattern, which matches the splitting pattern implemented by the transmitter’s stream splitter SST1 such that, after subsequent decoding, the corresponding original message as it arrived at the transmitter encoder ET1, is generated. The transmitter’s splitting order / pattern, - or at least an indication eg representation thereof - may be signalled from the transmitter to the receiver in a separate channel. Additionally or alternatively, the receiver may store, in embodiments permanently eg in ROM and / or fixed program code, that order (or indication) and / or the matching desplitting order / pattern. Regardless, the signalled and / or stored order is used to determine how the desired common part Sc,kA and estimated private part Sp,kA are combined to output the stream SkA from the receiver’s stream combiner SCR1. The decoder DR1 decodes the stream SkA from the receiver’s stream combiner SCR1, preferably to in effect reverse the encoding implemented at the transmitter. This may for example involve removing redundant bits (eg parity bits) added at the transmitter for FEC. The decoder DR1 may use a codebook that matches that applied by the transmitter encoder ET1 for an intended recipient that is the particular user of the receiver; same or different codebooks may be used by the transmitter and receiver for different message recipients. The decoder may for example be an LDPC decoder. However, and provided that the same channel coding scheme is used at the transmitter and receiver of an embodiment, other encoding I decoding techniques can alternatively be used, e.g., turbo-code or polar-code. In view of the above, decoding is notably not carried at an input, eg first, stage of the receiver. In CS-RSMA embodiments, decoding may even be a final stage before output of a reconstructed message to its intended recipient. Thus, at the transmitter side of the CS-RSMA system for multi-user unicast as shown in Fig. 3A, preferably all of the following steps are performed: 1) The transmitter first encodes the K unicast messages, {W1,..., WK}, into K codewords / data streams, {X-i,..., XK}. 2) Each codeword / data stream is segmented into common and private parts. For example, Xk is segmented intoX(c,k) andX(p,fc). 3) All the common parts, i.e. \X (c,1).....X (c,K)}, are combined into one common stream, Xc, while the K private parts remains as K private streams. 4) The K+^ data streams, {Xc, X(p,1),..., ^p.K")}, are modulated, precoded and transmitted. At the receiver side of the CS-RSMA system for multi-user unicast as shown in Fig. 3B, preferably all of the following steps are performed: 1) Each receiver obtains estimates, i.e., demodulated I demapped versions, of the common stream and the desired private stream. For example, the receiver Rl-k estimates sr and sn k as sr and sn k. 2) Each receiver then extracts from the common stream estimate its desired segment. For example, receiver Rl-k extracts sc,k from sc. 3) Each receiver combines estimates of the desired private stream and common stream segment into an estimate of a complete data stream. For example, receiverRl-k combines sc k and spk into sk. 4) Each receiver decodes the obtained data stream into the desired unicast message. For example, receiver Rl-k decodes sk into Wk. In view of the above, and notably, embodiments of the transmitter perform the splitting on encoded bits, eg after LDPC encoding. This contrasts with RSMA systems where the splitting is done with the uncoded message at the input stage of the transmitter architecture. Relatedly, in the disclosed CS-RSMA embodiments, the common stream as such does not need to be decodable (eg to reverse the transmitter LDPC encoding). Rather, each user / receiver may extract only a segment of the common stream, combine that with the private stream and then perform decoding; hence there is no need to directly decode the full common stream. (In embodiments, the full common stream generally cannot be decoded anyway because the common stream is not from a complete codeword). This contrasts with RSMA systems in which it is necessary to design the modulation and code rate of the common stream to be easily decodable by all relevant users. Thus, in examples of the transmitter, encoding eg LDPC is performed at an initial stage and modulation such as QAM, eg QPSK, is performed after the splitting and combining, eg immediately ahead of a precoding (eg using zero-forcing) stage. On the other hand, in examples of the receiver, the demodulation is performed by the estimator at an initial stage, whereas the eg LDPC decoding is performed after the decomposing and combining stages. For further detail, it is noted that in some embodiments signalling from the transmitter could be implemented to communicate to any receiver(s) what modulation technique, eg type of QAM or QPSK, is being applied to the common stream and / or the private stream desired to be received at the receiver. Additionally it is noted that, in any CS-RSMA embodiment, interleaving and / or scrambling may be implemented at the transmitter, and conversely de-interleaving and / or descrambling at a receiver. For example, and similarly as in the 5GNR standard, FEC encoding in a transmitter can comprise each of e.g. LDPC encoding, interleaving (of systematic and parity information) and scrambling. In an embodiment, the interleaving and scrambling could for example be applied just after the encoding or after the stream splitter and stream combiner of a transmitter. In view of the above, one or more of the following features / advantages may apply in a CS-RSMA embodiment: Splitting and combining in the transmitter are performed on the encoded codewords instead of original messages; Splitting and combining in the receiver are performed on the estimated codewords instead of original messages; and / or - The common stream does not have to be decodable to all the recipient users. In contrast to a particular user’s receiver of a known RSMA system, which needs to fully decode a common stream in a first, decoding unit of the receiver, an embodiment firstly estimates (rather than decodes) all received modulated symbols before the splitting and combining, which then provide user-specific data to the later decoding unit. Furthermore, and specifically regarding CS-RSMA for multi-user unicast, advantages may include one or more of: Low encoding / decoding requirements because, e.g.: the transmitter encodes (eg LDPC) K messages rather than K+1 messages as in known RSMA systems; the receiver decodes one codeword rather than two as in known RSMA systems; and / or any CRC checking requirements are reduced, etc.. Thus, embodiments allow complexity reduction at the transmitter and / or receiver. Reduced control signalling overhead because an embodiment does not need to signal to receiver users what is the coding scheme of the common stream. Thus, each receiver may only need to know the coding scheme for one codeword, rather than two (common and private) as with known RSMA. In embodiments, there is no need to signal to recipient users what is the coding scheme of a common stream, whereas known RSMA systems may need to communicate to a receiver a size (LPDC size) and code rate of the common stream; and / or Simplified re-transmission mechanism: there is a reduced need in embodiments for a receiver to send separate retransmission requests to a transmitter in the case of incorrect reception, eg when errors are detected at the receiver. The use of only k encodings in the transmitter means that a receiver may only send request retransmission requests for the k parallel pipelines, i.e., k retransmissions in respect of the k original incoming messages from the transmitter rather than k+1 retransmissions for the common stream and k private streams. Consequently, the overall re-transmission process may be simpler in embodiments. Specifically, for e.g. 5G NR (New Radio), an embodiment may only need to implement Hybrid Automatic Repeat Request (HARQ) for K codewords (same as for MU-MIMO); whereas in known RSMA systems, additional HARQ may need to be implemented for the common stream. Generally, embodiments may not require retransmission of a common stream because encodings are done only on the k original messages. It is to be additionally noted that, in the known RSMA systems, the HARQ process for common stream is different from that for the private: the HARQ for private streams can be the same as the HARQ used in other existing unicast systems (e.g. MU-MIMO in LTE / 5G), but HARQ for the common stream is more complicated due to the fact that the common stream needs to be decoded by multiple receivers. Hence, with known RSMA, efforts need to be made for common stream HARQ. In contrast, CS-RSMA embodiments can simply reuse the existing HARQ for unicast systems (e.g. MU-MIMO). For further detail regarding the above-mentioned retransmission requests, we note that embodiments may use different designs for the re-transmission process. In one example, a receiver or user reports that a message W1A was incorrectly received and would like a retransmission. The transmitter may then feed the message W1 again (or a corresponding redundancy version), while other unicast messages W2, W3, ....WK, would be new information messages. The re-transmission message W1 plus the new messages W2.....,WK would in this example then go through the process comprising the stream splitter, stream combiner, modulation and precoder. In this regard however, re-transmitting the same stream is a relatively basic example. For example, if in the first transmission only receiver / user 1’s decoding is erroneous, then in the next transmission, X1 will comprise the elements (eg bits) for user 1’s retransmission but, for other users, there will be new messages, W2, W3, etc. that lead to new X2, X3 streams and the subsequent process as before. In some embodiments, such as for 3GPP standards (e.g. LTE and 5G), a more efficient retransmission (HARQ) may be used, for example involving transmission of additional parity bits from LDPC codes. Further considering advantages, a CS-RSMA embodiment may additionally or alternatively have any one or more of the following: a) The change in the architecture to first encode messages into codewords / data streams and then split into common and private parts may have benefits in terms of transmitter and receiver complexity and / or requiring less overhead due to control signalling. b) In comparison to known MU-MIMO implementations, embodiments may enhance performance (e.g. throughput, latency, etc.) in different relevant scenarios such as high demand density areas (situations where a large number of users demand connectivity to a cell e.g., airports, sports venues, rail and subway stations and major public events). c) In comparison to known RSMA systems, embodiments may provide similar or improved performance (e.g. throughput, latency, etc.), while saving on transmitter and / or receiver (encoding and / or decoding) complexity and / or control signalling overhead, potentially simplifying a re-transmission mechanism. d) embodiments may be suitable as part of the global 3GPP standards, eg 4G (eg LTE) or 5G and / or even 6G, for communication between handsets and network equipment. Figs. 4A and 4B show a physical layer (PHY) transceiver of a CS-RSMA embodiment for unicast-only. In the transmitter of Fig. 4A: Wi - Wk are the uncoded bits intended for the K users; Xi - Xk are the coded bits intended for the K users; Xc,1 - Xc,K, Xc and Sc are common parts I segments from Xi - Xk; and Wp,1 - Wp,K and S1 - SK are private parts within the transmitter. In more detail, multiplexing Xc,1 - Xc,K leads to Xc that comprises the coded bits for the common stream, and modulating Xc and Xp,1 - Xp,K leads to Sc and S1 - SK, i.e. common and private symbol streams. Within the receiver of Fig. 4B, LLRc, and LLRc,k are common while LLRp.k is private (LLR denotes log-likelihood-ratio, as may be output from a de-modulator in embodiments; this ratio may be interpreted as a soft result from symbol detection, rather than the symbol itself). Thus, and regarding the Fig. 3A transmitter elements: the encoding may be implemented by the FEC blocks of Fig. 4A; the stream splitter may be implemented by the demultiplexer (DEMUX); the stream combiner may be implemented by the multiplexer (MUX); and / or the functions of Fig. 3A’s modulation and precoding blocks (the precoding optionally using zero forcing) may be implemented by Fig. 4A’s modulators M and precoding block. The modulators M are generally for modulation to constellation symbols such as QAM, eg, QPSK symbols; thus a stream S of Fig. 4A may comprise complex QAM values depending on what constellation order is used, eg QPSK / 16QAM. In the Fig. 4A transmitter embodiment, a resource mapping block is additionally provided in the output stage. Fig. 4B provides two receiver options: the upper one is a joint de-mapping receiver; and the lower one is a soft SLIC (symbol level interference cancellation) receiver. Regarding the Fig. 3B receiver elements relative to the upper PHY receiver embodiment in Fig. 4B,: the estimator may be implemented by the resource demapping and demodulator (M-1) blocks, the splitter by the demultiplexer, the combiner by the multiplexer (MUX); and / or the decoding by the reverse forward error correction (FEC-1) blocks. The demodulator M'1 may for example demodulate from QAM eg QPSK symbols of Yk, preferably by calculating distances of estimates from a received signal to determine what are the most likely symbols of the common and private streams. The lower PHY receiver embodiment in Fig. 4B has a corresponding structure as the upper embodiment, however the estimator comprises a second demodulator (M-1) to receive the resource demapping stream having subtracted therefrom a soft symbol unit output based on the output of the first demodulator. Regarding the soft symbol feedback as shown in Fig. 4B, the soft symbol block may in embodiments compute soft symbols (expectation of the symbols) from the LLRs. The soft symbol may then be subtracted from the received signal to make the estimation of private stream symbols. This may result - statistically - in less interference after the substraction and thus allow greater accuracy. Figs. 5A and 5B respectively show, again at the physical layer (PHY), another design of transmitter and alternative receiver configurations of a CS-RSMA unicast embodiment. Compared to Fig. 3A or 4A, at the transmitter of Fig. 5A, the order of the splitter and combiner stage (eg SST1 + SCT1, or DEMUX + MUX) and the modulation stage (M1, M) is reversed. As shown in Fig. 5B, no change is needed in the receiver relative to Fig. 3B or 4B. Fig. 9 shows link-level simulations of a unicast-only PHY transmitter I receiver CS-RSMA embodiment, compared to for a unicast-only PHY transmitter / receiver employing known RSMA. Simulation results are shown for a system using a joint demapping receiver and also for a system using a soft SLIC receiver; such receivers are shown respectively in the upper and lower diagrams of Fig. 4B. Specifically, the simulations are based on Modulation and Coding Schemes (MSC) involving Quadrature phase-shift keying (QPSK): CS-RSMA using QPSK+0.5; and known RSMA using QPSK+0.85(common) and QPSK+0.325 (private). (For clarity: QPSK+0.5 refers to code rates, e.g. QPSK+0.5 means the bits are first encoded (using LDPC for example) at a code rate of 0.5, then modulated using QPSK constellation). Figs. 6A and 6B show a CS-RSMA embodiment for joint unicast and non-unicast (i e., multicast and / or broadcast). The functions and characteristics of the elements of Figs. 6A and 6B are generally as described above for corresponding elements of Figs. 3A and 3B (each figure pair correspondingly comprising: encoding, stream splitter, stream combiner, modulation and precoding in the transmitter; and estimator, stream decomposer, stream combiner and decoding in the receiver). Therefore, generally differences relative to Figs. 3A and 3B are discussed in the following. Merely for assistance in this regard, we note that unicast involves each of multiple users receiving independent information (e.g., where each user is independently messaging or requesting social media content), whereas multicast and broadcast involve the same content being received by multiple users (all users in the case of broadcast); multicast and broadcast may be used for example for distributing a live stream of a sporting event or for social media distribution. As shown in Fig. 6A, the transmitter T2 receives both a non-unicast (multicast or broadcast) message Wm and K unicast messages W1 - WK. Following encoding of the unicast and multicast / broadcast messages, and splitting by a stream splitter SST2 of the K unicast messages into K common parts (Xc,1 - Xc,K) and K private parts (Xp,1 - Xp,K) according to a splitting order / pattern as described for Fig. 3A, the K common parts are input together with the encoded multicast / broadcast stream Xm to a stream combiner SCT2, which in turn outputs a combined unicast and multicast / broadcast common stream Xc to a modulation stage M2 (preferably for QAM eg QPSK modulation). The modulation results are forwarded to the precoder such that the split and modulated private parts are separately input to the precoder. The combining of the unicast and multicast / broadcast common streams in the stream combiner SCT2 is performed based on a combining order / pattern applied to concatenate / append / interleave etc, similarly again as described for the stream combiner SCT1 of Fig. 3A. The combining order may involve concatenating, interleaving, and / or applying logic operation(s) to respective users’ unicast common parts and the multicast / broadcast common stream, to build the combined common stream Xc. However, any order, eg combining pattern, can be implemented provided that a matching decomposing operation is performed at each receiver. Each receiver R2-k of Fig. 6B receives signals respectively corresponding to the transmitted common stream and desired private part (i.e., the signals that are receivable at the particular receiver dependent on the type of multiplexing, eg OFDM, implemented at the transmitter). Correspondingly to the transmitter T2, an estimator ER2 estimates the symbols of the common stream and the desired private part. The stream decomposer then decomposes the estimated common stream ScA, to separately output the estimated multicast / broadcast stream Sma (eg live TV program) and the estimated unicast common part Sc,kA corresponding to the particular receiver / user. Generally, the splitting is based on a decomposing order / pattern that matches the order in which basis the transmitter stream combiner SCT2 combined the multicast / broadcast stream and the common parts of respective (encoded) user messages to create the common stream Xc. That order, eg combining pattern, - or at least an indication thereof - may be signalled from the transmitter T2 to the receiver R2 in a separate channel. Additionally or alternatively, the receiver may store, in embodiments permanently eg in ROM and / or fixed program code, that order (or indication) and / or the matching decomposing order / pattern. Regardless, the signalled and / or stored order is used to determine how the desired common part Sc,kA and multicast / broadcast stream Sma are separated from the estimated common stream ScA for output by the stream decomposer SDR2. The stream combiner SCR2 then combines the unicast common part Sc,kA and the unicast private part Sp,kA corresponding to the receiver user, to output the user’s unicast stream SkA for decoding. Similarly as for Fig. 3B, the combining may be based on a desplitting order or pattern, which matches the splitting pattern implemented by the transmitter’s stream splitter SST2 such that, after subsequent decoding, the corresponding original unicast message as it arrived at the transmitter encoder, is generated. The transmitter’s splitting order / pattern, - or at least an indication thereof - may be signalled from the transmitter to the receiver in a separate channel. Additionally or alternatively, the receiver may store, in embodiments permanently eg in ROM and / or fixed program code, that order (or indication) and / or the matching desplitting order / pattern. Regardless, the signalled and / or stored order is used to determine how the desired common part Sc,kA and estimated private part Sp,kA are combined to output the stream SkA from the receiver’s stream combiner SCR2. The decoder D2 may perform two decodings, preferably using the same type of decoding (eg to remove redundancy added by LDCP encoding at the transmitter): one to generate the multicast / broadcast stream Wma and another for the unicast stream WkA. In comparison to known RSMA systems for joint unicast and non-unicast, a unicast and non-unicast (multicast and / or broadcast) embodiment such as that of the Figs. 6A and 6B systems may exhibit one or more of the following: Same number of messages to be encoded at the transmitter and decoded at the receivers. This may be the case because, in an embodiment, the transmitter will also encode the non-unicast, e.g., multicast, input stream; Same amount / number of control signals / signalling for modulation and coding scheme. This may be the case where, in a joint unicast and non-unicast embodiment, the transmitter signals to the receiver what is the coding scheme used for the common stream; and / or - Simplified re-transmission mechanism (e.g. in the case that unicast messages are correctly decoded but the non-unicast message is not, a preferred CS-RSMA embodiment may only need to follow eg the HARQ process for the previous nonunicast message for example as in 5G NR (e.g. transmit redundancy versions stored in the buffer); known RSMA systems may combine the previous nonunicast message with new unicast messages and perform encoding again (i.e. HARQ may not be applied to multicast / broadcast). A re-transmission scheme suitable for an embodiment such as shown in Figs. 6A and 6B does not need to be re-designed: known schemes - such as HARQ and retransmission schemes used in 5GNR that includes multicast or broadcast capability - remain applicable, even though the common stream is intended for different users. Figs. 7A and 7B show, at the physical layer (PHY), an embodiment of a transceiver of CS-RSMA for joint unicast and non-unicast (i.e., multicast and / or broadcast, wherein Wm, Xm and LLRm are respectively the uncoded bits, coded bits, and LLRs for the non-unicast (multicast / broadcast) message. Similarly to Figs. 4A and 4B): Xc,1 - Xc,k, Xc and Sc are common parts, whereas Xp,1 - Xp,k and s1 - sk are private parts within the transmitter and, within the receiver, LLRc, LLRc,k and LLRm are common parts while LLRp,k is a private part. Thus, and correspondingly to the Fig. 6A transmitter elements: the encoding may be implemented by the FEC blocks of Fig. 7A; the stream splitter may be implemented by the demultiplexer (DEMUX); the stream combiner may be implemented by the multiplexer (MUX); and / or the functions of Fig. 6A’s modulation stage and subsequent precoding block (eg using zero forcing) may be implemented by Fig. 7A’s modulators M and precoding block. Additionally, Fig. 7A shows resource mapping blocks similarly as described for Fig. 4A. (As above in relation to unicast-only, in a CS-RSMA unicast and non-unicast (multicast and / or broadcast) system, the receiver will have a resource de-mapping block if resource mapping is done at the transmitter). Similarly as for Fig. 4B, Fig. 7B provides two receiver options: the upper one is a joint de-mapping receiver; and the lower one is a soft SLIC (symbol level interference cancellation) receiver. Regarding the Fig. 6B receiver elements relative to the upper PHY receiver embodiment in Fig. 7B,: the estimator may be implemented by the resource de-mapping and / or demodulator (M’1; such as for demodulation from QAM eg QPSK) blocks; the splitter by the demultiplexers; the combiner by the multiplexers (MUX); and / or the decoding by the reverse forward error correction (FEC-1) blocks. The demodulator may estimate the most likely transmitted symbols based on estimating distances. Again similarly to Fig. 4B, the lower PHY receiver embodiment in Fig. 7B has a generally corresponding structure as the upper receiver embodiment, however a second demodulator (M’1) receives the resource demapping stream having subtracted therefrom a soft symbol unit output based on the output of the first demodulator. Similarly as for Fig. 4B, the soft symbol block may in embodiments compute soft symbols (expectation of the symbols) from the LLRs. The soft symbol may then be subtracted from the received signal to make the estimation of private stream symbols. This may result - statistically - in less interference after the substraction and thus allow greater accuracy. Figs. 8A and 8B respectively show, again at the PHY layer, another design of a transmitter and alternative receiver configurations of an embodiment of CS-RSMA for multi-user unicast and non-unicast (multicast and / or broadcast), eg NOUMB. Compared to Figs. 6A and 7A, at the transmitter of Fig. 8A, the order of the splitter and combiner stage (eg SST2 + SCT2, or DEMUX + MUX) and the modulation stage (M) is reversed. As shown in Fig. 8B, no change is needed in the receiver relative to Figs. 6B or 7B. More generally regarding any transmitter embodiment described herein, the splitter / splitting - or any demultiplexer used for that function - may split bits / symbols into two segments (in contrast, for generalized or hierarchical RS it can be more than two segments). The number of bits / symbols in each segment can be arbitrarily determined. In extreme CS-RSMA embodiments, one of the segments may contain all the bits / symbols and the other segment may contain no bit / symbol. In such embodiments, this may lead to scenarios where the common stream effectively contains only the coded bits / symbols from a subset of receivers, or some of the private streams are deactivated as there is nothing to transmit. Regarding the above-mentioned determination of the number of bits / symbols, we further note that in any transmitter embodiment, the splitter for each encoded message may have the flexibility to allocate from 0 to 100% of the bits / symbols to the common part corresponding to that message. For example: if 100% of the bits / symbols are allocated to the common part then there is no private stream for that user; and if 100% of the bits are allocated to the private part this means that no bits from that user are allocated to the common stream. Thus, and using the notation applied in the drawings, if the splitter of an encoded message allocates all the bits / symbols to the common stream, then Xc,1 = X1. In a more specific example where there are K=2 users, embodiment may split the information from user 1 into a common and private message and may not split information from user 2. In this case, Xc,2 is empty, Xc = Xc,1 and the resulting output streams are: Sc (only carrying information from useri), S1 (carrying some information from useri) and S2 (carrying all the information from user 2). In other words, a transmitter embodiment will generally split at least one user’s encoded message(s), e.g., 1 or all, so that each such split message contributes to the common stream and also have a private part. However, embodiments may not split all encoded messages. In such an embodiment, in addition to the splitting of at least one user’s encoded messages,: encoded messages from one or more users are unsplit, such that they don’t contribute to the common stream and effectively go through as if they are private only; and / or encoded messages from one or more users are unsplit, such that they don’t contribute to the private streams and effectively go through as if they are common only. To achieve the above, a splitter may be controlled to allocate a portion, e.g., 0 to 100%, of each encoded message of a user to the common stream. For example, a base station or other apparatus comprising the transmitter may determine the splitting pattern for each user, depending on channel state information that users may send to the transmitter and / or on service requirements. These splitting pattern(s) would then be informed to each user equipment (UE) having a receiver embodiment, the informing using signalling as described herein. Consistent with the above, and in any embodiment described herein, the same splitting pattern may be applied for all transmitter users, or different splitting patterns may be applied for at least one of respective users. These patterns(s) are fixed within a system or communicated to receivers, so that the appropriate pattern is known at each receiver. Merely for completeness, Fig. 10 shows an example computing device or system on which signal processing of an embodiment may be implemented. For example, computing device / system comprises a bus, processor(s), communication port(s) (e.g., RS232, Ethernet, USB, etc.), and / or memory, all generally coupled by a bus (e.g., PCI, SCSI). Each element of Fig. 10 is optional. The memory may comprise non-volatile memory such as read only memory (ROM) or a hard disk and / or volatile memory such as random access memory (RAM, e.g., SRAM or DRAM), cache (generally RAM) and / or removable memory (e.g., EEPROM or flash memory). The processor may be any known processor, e.g., an Intel (registered trademark) or ARM (registered trademark) processor. A user interface, e.g., display screen and / or keyboard may be provided. We further note that Figs. 6A, 7A and 8A use the subscript M, for example in relation a single incoming multicast message Wm. In embodiments, such a multicast message may be replaced by multiple multicast messages and / or at least one broadcast message. No doubt many other effective alternatives will occur to the skilled person. It will be understood that the invention is not limited to the described embodiments and encompasses modifications apparent to those skilled in the art lying within the spirit and scope of the claims appended hereto.

Claims

1. A transmitter comprising:an encoder to encode incoming messages from respective said users of the transmitter;a stream splitter to split each encoded message to form from each split message a common part and a private part, each of the parts having a number of elements such as bits or symbols, wherein each said splitting of a said encoded message is based on a splitting order that determines elements of the encoded message that are to form the common part of that encoded message and elements of the encoded message that are to form the private part of that encoded message, wherein the number of elements of each of the common part and the private part of at least one of the split encoded messages is more than or equal to one;a stream combiner to form a common stream comprising each said formed common part of the encoded messages, wherein the forming is based on a combining order that determines where each element of each common part is positioned in the common stream; andan output stage for outputting signals for transmission based on the common stream and private parts.

2. The transmitter of claim 1 for unicast communication, wherein the incoming messages are unicast messages.

3. The transmitter of claim 1 for unicast and non-unicast communication, wherein the incoming messages comprise unicast messages and the transmitter is further for transmitting a non-unicast message, wherein non-unicast is multicast or broadcast, and wherein:the encoder is to encode the unicast and non-unicast messages; andthe stream combiner is to combine the encoded non-unicast message and each said formed common part of the encoded unicast messages to form the common stream, wherein the combining is based on a combining order that determines where each element of each common part and of the encoded nonunicast message is positioned in the common stream.

4. The transmitter of any preceding claim, comprising:a modulation stage to modulate the common stream and the private parts and forward the modulated stream and parts to the output stage, wherein the outputting is based on the forwarded stream and parts,preferably wherein the output stage comprises a precoder to precode signals based on the modulated stream and parts and to form the outputted signals based on the precoded signals.

5. The transmitter of any one of claims 1 to 3, comprising:a modulation stage to modulate encoded unicast messages from the encoder and forward each modulated unicast message to the stream splitter, wherein the splitting splits the forwarded messages to form the common and private parts, preferably wherein the transmitter is according to claim 3 and the modulation stage is further to modulate an encoded non-unicast messsage from the encoder and forward the modulated non-unicast message to the stream combiner,preferably wherein the output stage comprises a precoder to precode signals based on the formed stream and parts, wherein the outputted signals for transmission are based on the precoded signals.

6. The transmitter of any preceding claim to, in response to receiving a message retransmission request, retransmit a said common stream and private part having elements of a message corresponding to the message retransmission request,preferably wherein the request and response are in accordance with a HARQ process.

7. A receiver comprising:an estimator to generate modulation symbol detection results of a received common signal and to output the results as an estimated common stream, and to generate modulation symbol detection results of a received private signal and to output the results as an estimated private part;a stream decomposer to extract from the estimated common stream results to form an estimated common part corresponding to the receiver user, wherein the extracting is based on a decomposing order; anda stream combiner to combine the estimated common part corresponding to the receiver user and the estimated private part to generate an estimated message stream, wherein the combining is based on a desplitting order.

8. The receiver of claim 7 for unicast communication, wherein the estimated message stream comprises an estimated unicast message.

9. The receiver of claim 7 or 8 for unicast and non-unicast communication, wherein non-unicast is multicast or broadcast, wherein:the extraction by the stream decomposer further extracts from the estimated common stream results to form an estimated non-unicast stream, wherein the extraction to form the estimated common part and the estimated non-unicast stream is based on the decomposing order.

10. The receiver of any one of claims 7 to 9, comprising a decoder to decode the generated estimated message stream and to output a message based on thedecoded stream, preferably wherein the decoder performs FEC such as LDPC decoding.

11. The receiver of claim 10, wherein the decoder is further to decode the estimated non-unicast stream and to output a non-unicast message based on the decoded estimated non-unicast stream.

12. The receiver of claim 10 or 11, wherein the decoder is to detect an error in estimated stream corresponding to a message, andwherein the receiver is to, in response to a said error detection, transmit a request for retransmission of a common stream and private part corresponding to the message, the retransmission request preferably in accordance with a HARQ process.

13. A communication system, comprising the transmitter of any one of claims 1 to 6 and at least one receiver of any one of claims 7 to 12.

14. The communication system of claim 13, wherein the decomposing order of the receiver matches the combining order of the transmitter.

15. The communication system of claim 13 or 14, configured to communicate to the receiver an indication of the combining order of the transmitter.

16. The communication system of any one of claims 13 to 15, wherein the desplitting order of the receiver matches a splitting order of the transmitter.

17. The communication system of claim 16, configured to communicate to the receiver an indication of a splitting order of the transmitter.

18. The communication system of any one of claims 13 to 17, wherein the transmitter and receiver are to perform a HARQ process.

19. The communications system of any one of claims 13 to 18, wherein the system is a MU-MIMO system, such as a massive MIMO system,preferably wherein the system is a multi-layer MIMO system such as for 4GLTE and 5GNR, configured to map the signal to be outputted for transmission based on the common stream and the signals to be outputted for transmission based on the private parts to multiple MIMO layers.

20. The communications system of any one of claims 13 to 19, for use in a cellular telecommunications, wifi or satellite communications network, such as a 5G NR network.21.The transmitter, receiver or communications system of any preceding claim, for RSMA such as cooperative rate splitting RSMA, hierarchical rate splitting RSMA or multilayer splitting RSMA.

22. A transmission method comprising:encoding incoming messages from respective users into codewords; andsplitting at least one of the encoded codewords to form from each split codeword a common part and a private part, wherein each said splitting of a codeword is performed according to a splitting order, wherein the common part and the private part formed from at least one encoded codeword each comprise one or more elements of the codeword;forming a common stream comprising each split common part of the at least one of the encoded codewords, wherein the forming comprises combining elements of the common parts according to a combining order;outputting signals for transmission based on the common stream and private parts.

23. The transmission method of claim 22, for unicast and non-unicast communication, wherein the incoming messages comprise at least one unicast message and a non-unicast message, wherein non-unicast is multicast or broadcast, wherein:the encoding encodes the unicast and non-unicast messages; andthe combining combines the common parts of the encoded unicast messages and the encoded non-unicast message to form the common stream, wherein the combining is based on a combining order that determines where each element, of each common part and of the encoded non-unicast message is positioned in the common stream, preferably wherein the elements are either bits or symbols.

24. A receiving method comprising:receiving a common signal and a private signal;generating modulation symbol detection results of the common signal and outputting an estimated common stream based on the results;generating modulation symbol detection results of the private signal and outputting an estimated private part based on the results;extracting from the estimated common stream a common part for a respective user, the extraction based on a decomposing order;combining the estimated private part and the extracted common part to generate a message stream, the combining based on a desplitting order; and thendecoding the generated message stream.

25. The receiving method of claim 24, for unicast and non-unicast communication, wherein non-unicast is multicast or broadcast, wherein:the extracting further extracts from the estimated common stream symbols to form an estimated non-unicast stream, wherein the extraction to form the estimated common part and the estimated non-unicast stream is based on the decomposing order; andthe decoding decodes the generated message stream to output a message, and decodes the estimated non-unicast stream to output a non-unicast message.

26. A communication method comprising the transmission method of claim 22 and the receiving method of claim 24, or the transmission method of claim 23 and the receiving method of claim 25, wherein the signals transmitted based on the common stream and the transmitted signals based on the private parts are the received common and private signals, respectively.

27. A communication method of any one of claims 22 to 26, comprising the transmission method of claim 23 and the receiving method of claim 25, wherein the transmitter signals to the receiver a type of coding scheme used for the received common signal.

28. A data carrier carrying processor control code which when running on a processor causes the processor to implement the method of any one of claims 22 to 27.38

Images