DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

[0044] The methods and embodiments of the network modem disclosed herein enable implementations of advanced communications providing with high quality real-time processing with interferers. The preferred embodiment is just an illustration of the apparatus and processes that are applicable to many variations and applications all within the scope of the invention.

[0045] FIG. 1 shows an example network 10 of terminals 12 - 15 communicating through a satellite transponder 16 . The satellite transponder 16 is one embodiment of a signal transponding element, and other elements performing a similar function include repeater towers for terrestrial point-to-point radio and microwave communications and fixed transponders for local area networks. While the physical means differ, the network architecture necessary to employ the present apparatus is the same. It is also within the scope of the invention that the terminals communicate directly without transmitting through a transponding element.

[0046] In each case, user terminals 12 - 15 communicate pair-wise with the other terminals using bi-directional traffic channels 20 - 23 radiating through the transponding element as shown in FIG. 1 . These channels 20 - 23 may be allocated according to separate frequencies in a frequency domain multiple access scheme (FDMA), time slots in a time domain multiple access scheme (TDMA), codes in a code division multiple access scheme (CDMA) or any other means of partitioning signals, including the hybrid forms.

[0047] Both the bandwidth and propagation conditions differ for the different terminals 12 - 15 in the network 10 . In the example of FIG. 1 , terminal 12 suffers from poor propagation due to the presence of clouds 25 , while the other terminals can have excess signal margin due to clear line of sight propagation conditions. As known to those skilled in the art, there are many factors that can affect signal communications and clouds 25 are just a single factor.

[0048] A network control terminal 15 allocates traffic communication channels to the other terminals 12 - 14 . The network control terminal 15 communicates channel assignments using a signaling channel, which in this embodiment is within the bi-directional traffic channel 23 . The signaling channel of the bi-directional traffic channel 23 is used for signaling channel assignments, status messages and other activity necessary to maintain inter-terminal communications on the traffic channels. This signaling channel 23 can be a dedicated link between the controlling element and each user terminal or it can be a shared channel. In cases where it is a dedicated link, its bandwidth is generally much lower than the traffic channel bandwidth as signaling is relatively infrequent and low throughput compared to data traffic. It is also possible to implement the signaling channel as a shared channel using an ALOHA protocol or carrier sense multiple access protocol or some other shared channel protocol.

[0049] The network 10 is heterogeneous in the sense that some terminals in the network use conventional, legacy modems while others use the bandwidth-efficient modems described herein. Channel communication may be direct between terminals via the bi-directional channels 20 - 23 as shown in FIG. 1 . It can also be indirect through a hub element, or directly via a hardwire interconnection. The network channel communication capability is important, while the means for implementing arises in many forms.

[0050] In one embodiment each of the bandwidth efficient modems are capable of operating as the network control terminal. The internal hardware elements and microprocessor of any of the terminals can perform as the primary network controller to coordinate communications. The selection of the network controller terminal can be based on whichever terminal has the best signal power and communication ability with the other transponder or other terminals. If the present network terminal is blocked or malfunctioning, another terminal can take control.

[0051] The network control terminal 15 monitors the status of each terminal 12 - 14 in the network to ensure that each user is able to operate as close as possible to its desired data throughput rate. The network control terminal 15 monitors the signal quality in each channel 12 - 14 as reported in terms of bit error rate communicated by all terminals in the network 10 . One or more of the terminals 12 - 14 can be a bandwidth efficient modem, which also reports signal to noise (S/N) and signal to interference (S/I) levels. The network control terminal 15 assigns traffic channels in a manner that exploits the capability of bandwidth efficient modems to demodulate signals in the presence of strong interference.

[0052] Specifically, when sufficient signal-to-noise margin exists, the network control terminal 15 will assign terminals with bandwidth efficient modems to channels already occupied by other terminals. In cases where the channel is occupied by legacy terminals, the signal level will be set sufficiently low to fall within carrier-to-interference level requirements for the legacy terminal. In cases where the channel is occupied by a bandwidth-efficient terminal, the signal level will be set higher, but still low enough for both terminals to demodulate their signals even with the other modem occupying the same channel.

[0053] FIG. 2 shows the overall architecture of the bandwidth-efficient modem 50 . Signals come in through the receive antenna 52 into a front end 54 . The signals arriving at the antennas 52 , 74 contain multiple users, multipaths, and other interfering signals. There can be separate reception antennas 52 and transmitting antennas 74 , or a single antenna can be used with a diplexer to switch between reception and transmission. In certain applications the antenna(s) 52 , 74 are an integral part of the network modem assembly, while other applications can employ external antennas or even no antennas, if the signals are transmitted in a non-wireless application such as a cable modem.

[0054] The front end 54 generally performs amplification and downconversion and typically comprises an amplifier, filter and a mixer to downconvert the RF signals to baseband signals, with a digital to analog (D/A) converter to digitize the waveform. The downconversion removes some or all of the carrier component, leaving the baseband signals. Alternatively, with the high computer processing speeds, the received signal could simply be processed through an A/D converter without any RF components. The functionality and design of the front end units are well established to those familiar with the art.

[0055] The output of this front end receiver 54 is a baseband waveform 56 , which serves as input to the baseband processor 58 . It is a bit waveform representation of the RF signal generated by digitized sampling.

[0056] Baseband processor 58 employs digital signal processing in a manner described herein to separate the signals transmitted simultaneously by different transmitters over the same communication channel. The baseband processor output 60 produces its data from the intended transmitter on the interfering channel, signaling data from the signaling channel 23 and information about the interfering signals on the respective traffic channel. The message and signaling data are necessary for operation of the modem in a manner similar to conventional, legacy modems not designed to operate in the presence of interference. The data about the interfering signals is also necessary to operate in a saturated or overloaded environment where multiple communicators simultaneously transmit on the same channel. There is a large body of work involving signal cancellation methodologies that can be implemented in conjunction with the present invention.

[0057] Specifically, the baseband processor 58 calculates the power levels of every interfering signal on the channel, the channel thermal noise level and estimated bit error rates for each signal in the channel. Periodically, this data is transmitted over the signaling channel 23 to the network controller 15 shown in FIG. 1 .

[0058] The periodic transmissions to the network controller 15 can be based upon request, time, or other signal occurrences. The request, for example, would be from the controller 15 to each of the bandwidth efficient modems or through a broadcast message. The time interval for the periodic transmission of the modem data can be a fixed predetermined time period or a random time period. Finally, the periodic transmission could be based upon certain measurements that trigger the data transmission. The network controller 15 uses this information to make channel assignments in a manner consistent with efficient utilization of allocated bandwidth. The bit processing computer 70 shown in the figure performs error correction, descrambling and packet formatting as is done in conventional modem.

[0059] FIG. 2 also shows a data transmission path for the modem to send its data back into the network. This data consists of message information and signaling information wherein the bits are formatted, scrambled and encoded in the bit processor 70 as inputs to the waveform generator 72 . The waveform generator 72 creates a baseband modulated digital signal in a manner familiar to those practicing the art. The modulation can be binary phase shift key, quadrature phase shift key, Gaussian minimum shift key, quadrature amplitude modulation or other well-known digital modulations. This baseband waveform is upconverted to the carrier frequency and amplified in front end 54 , and transmitted through the transmit antenna 74 . It is within the scope of the present invention to use a single antenna for transmission and reception by time multiplexing the transmission and reception of information over the channel.

[0060] FIG. 3 shows the baseband waveform processor 58 in greater detail. This processor 58 performs digital signal processing using ASIC's, FPGA's, programmable digital signal processing chips or microprocessors as the data rates of the channel necessitate, wherein the more general term of digital signal processor will be used to refer to any of the processing types. As inputs, the interfering signals on the channel are fed into the synchronizing element 80 . This synchronizing module 80 is sometimes called a parameter estimator and typically estimates the amplitudes, carrier phases, carrier frequencies, symbol rates and symbol phases for each interfering signal in the communication link. It estimates and tracks these parameters continually while the modem is receiving data, and there is a training and acquisition phase similar to a phase lock loop. There are many parameter estimator units, and one embodiment is described in the published U.S. patent US 2002/0037061 A1 entitled System for Parameter Estimation and Tracking of Interfering Digitally Modulated Signals, which is incorporated by reference for all purposes.

[0061] In order to maintain accurate estimates of all interfering signal parameters, the synchronizing element 80 uses the demodulated output from the joint detector 82 . The joint detector 82 uses the parameters estimated by the synchronizing element 80 and the input interfering signals to demodulate the interfering signals by estimating the values of the bits transmitted by each interfering signal over the channel. The joint detector 82 tracks the various parameters and creates signal constellations that estimate symbols for each interfering signal.

[0062] The joint detector 82 demodulation can utilize multi-user detection (MUD) technology or any other suitable form of demodulation to separate signals. Multi-user detection (MUD) refers to the detection of data in non-orthogonal multiplexes. A MUD receiver jointly demodulates co-channel interfering digital signals, and there are various multiuser detectors in the prior art, including optimal or maximum likelihood MUD, maximum likelihood sequence estimator for multiple interfering users, successive interference cancellation, TurboMUD or iterative MUD, and various linear algebra based multi-user detectors such as all of those detailed in the well-known text “Multiuser Detection” by Sergio Verdu.

[0063] These demodulated values are output to the bit processing computer 70 of FIG. 2 . In addition to producing estimates of the bit values, the joint detector 82 also produces estimates of the bit error rate for each signal in the channel. The method of determining the bit values and the bit error rate is described in further detail herein. The bit estimates are also fed back to the synchronizing element.

[0064] FIG. 4 shows how the bit processing computer 70 estimates the power level of each interfering signal in the channel and the residual random thermal noise power. Using the bit value estimates from the joint detector 80 and parameter estimates from the synchronizing element 82 , the bit processing computer 70 reconstructs an estimate of each interfering signal. This processing requires knowing the power of every interfering element in the channel whether signals or noise because the scheduler of the controller terminal, such as in a satellite terminal or wireless LAN terminal, depend on the quality of service of each element experience. The bit processor 70 calculates the signal to noise ration (SNR) and carrier to interference ratio (CIR) for every communicating element in the channel.

[0065] The calculation requires adding all the signal models and comparing the total signal power in the channel to the total power in the channel and calculating the signal noise. The noise power is required to determine whether there is sufficient margin in the channel to add signals onto each other.

[0066] Signal reconstruction is performed by generating signal models using the Iamplitudes and phases from the parameter estimation and the bits from the baseband waveform processor. The signal models are generated according to the formula:

A _k e ^jphi(t) m ( t,Sphi,b )

[0067] where

[0068] A is the signal amplitude,

[0069] phi is the carrier phase and frequency offset,

[0070] b is the estimated bit stream and

[0071] Sphi is the symbol phase, and

[0072] m denotes the modulating function such as BPSK, QPSK or other digital modulations.

[0073] This function also includes the symbol pulse shape that is known for all modems in the system as part of the system design specification. These model signals are also termed remodulated signals.

[0074] Referring to FIG. 4 , the linear algebraic processing is accomplished as follows:

[0075] X=S+N, wherein

[0076] X=received signal;

[0077] S=transmitted information signals; and

[0078] n=thermal noise

[0079] n=X−S 1 $P_S={S}^2=\sum _{m=1}^NS_m^2$

[0080] wherein Ps represents the norm of the vector S, and N is the number of samples. This is the total received signal power and is the signal component of the total received signal X. The individual signals 80 are summed by the summer 85 for each of the N signal samples, and processed to generate Ps.

[0081] And,

P _n =∥n∥=∥X−S∥

[0082] wherein Pn represents the norm of vector n, and is calculated according to FIG. 4 by subtracting 87 the total received signal X from the summed signal and generating P _n . The SNR is the ratio of P _s /P _n .

[0083] The thermal noise is the random component of noise from the environment and receivers, such as from resistors in the receiver. There are other elements, including some forms of noise that do have some predictable attributes, and any received component that has some structured form can be processed.

[0084] Each amplitude estimate from the synchronizing element 80 yields a carrier-to-interference ratio (CIR) estimate according to the formula:

CIR _k =|A _k | ² /(| S| ² −|A _k | ² )

[0085] where A _k is the estimate of signal k's amplitude and S is the mean amplitude of the total received signal.

[0086] The residual noise signal is found by sequentially subtracting each interfering signal model from the baseband waveform. When the model signals are subtracted from the baseband waveform, the channel SNR is found by

SNR _k =|A _k | ² /(| P _k | ² )

[0087] where P _k is the mean amplitude of the residual signal. SNR is an important measure for determining the available capacity of the channel for signal k.

[0088] While interfering signals can be removed due to their deterministic modulation structure, the some elements of residual thermal noise, such as white Guassian noise, are random and unpredictable and cannot be estimated and removed from the channel. However, any noise elements that can be statistically estimated or determined, can be filtered and result in improved performance.

[0089] Table A lists the messages and parameters necessary for operating the control channel signaling interface controller, which exists inside the modem bit processing computer 70 . It communicates with the network controller in order to allocate channels among the communicators in an optimal manner. The messages fall into three categories. Channel status parameters are used to estimate the current excess capacity of the channel. These parameters include the carrier-to-interference levels in the channel, the signal-to-thermal noise level and the actual bit error rates. Channel service request messages are used to transact the assignment and to release channels among the user terminals. These messages are initiated by the user terminal and sent to the network controller. Network messages sent to the terminal include status queries, channel reassignments and power changes. 1

[0090] FIG. 5 shows the modem controller 90 . Its functions may be implemented as ASIC's, FPGA's, microprocessors, DSP's or software modules in a programmable processor or as some combination of these elements. The preferred embodiment will depend on the channel message rate and the state of the art in the components at the time of implementation. Input bit streams are error-corrected in the joint detector 82 before transmission into the modem controller 90 . The error-corrected bit sequence is descrambed in the descrambler 92 . The descrambled data is then frame formatted in the frame processor 94 and demultiplexed, producing the message bits from the user terminal.

[0091] The interface processor 10 extracts control signals from the input data stream and decodes control messages. The messages are shown in Table A. The interface processor 100 may transmit messages in response to received messages as shown. The control signal interface 102 selects data channels based on interface processor inputs.

[0092] The control signal interface processor 102 transmits the interference data to the network control element through the signaling channel. It configures the modem transmit and receive channels

[0093] The modem controller 90 also contains channel selectors 96 , 104 for selecting the correct input channel and for multiplexing the transmit bits onto the correct output channel. The modem controller 90 formats the data into the correct frame or packet structure with appropriate header bits and tail bits. It then optionally error encodes and scrambles the data to be consistent with the communications protocol. The formatted bit sequence is input to the waveform generator described in FIG. 2 for digital modulation.

[0094] Table B is a table summarizing the operation of the network control element. The control element assigns channels according to four network modes. ‘Applique’ radios contain the improved demodulator shown in FIG. 3 . ‘Legacy’ radios are network radios without the improvement. The present invention provides incremental improvement in network performance as improved bandwidth-efficient radios are added to the network. It is not necessary to replace all of the legacy radios in the network to improve bandwidth utilization. As more legacy radios are replaced overall network capacity will continue to increase. 2

[0095] The modes are determined by periodically polling the network elements for their calculated interference parameters. The interference parameters give a measure of the Quality of Service provided in the channel. There are four modes addressing different levels of aggregate demand for network service.

[0096] 1. In clear' mode, each terminal has a dedicated channel, enjoys good propagation and is communicating at full capacity with light encoding. This mode is used by conventional wireless networks.

[0097] 2. In ‘dense’ mode, there is more demand than clear channel capacity allows. The network controller allocates additional bandwidth by identifying bandwidth-efficient modems operating in the network and assigning the bandwidth-efficient modems pairwise to the same channel. For example, if A, B, C and D are user terminals employing bandwidth-efficient modems, the communication channels for “A transmitting to B” and “B transmitting to A” are simultaneously assigned as communication channels from “C transmitting to D” and “D transmitting to C” respectively.

[0098] 3. In ‘saturated’ mode, all bandwidth-efficient modem have been packed into communication channels with two, three or more modems simultaneously transmitting on the same channel as propagation and SNR conditions will allow. Therefore, the network controller begins assigning bandwidth-efficient modems to channels occupied by legacy modems not capable of demodulating signals in the presence of strong interference. These assignments can be made in cases where legacy radios have sufficient downlink margin due to good siting and propagation to permit operation of bandwidth efficient modems sufficiently high above the noise power to permit signal recovery, yet at sufficiently low power as to not interfere with the legacy demodulators. Example power levels are shown in Table D.

[0099] 4. In ‘degraded’ mode, the network controller signals to bandwidth-efficient modems to increase coding redundancy. This mode occurs when too many terminals attempt to communicate for the channel bandwidth and propagation available. Links are maintained on all of the channels at lower throughput using a technique to be described in a future patent.

[0100] Table C summarizes the channel assignment algorithm employed by the network controller. This algorithm is executed periodically by the channel scheduler. One implementation would be to run the scan once per superframe, that is, per set of frames such that each terminals has one control slot for communicating status, channel scheduling and control. 3

[0101] The network scheduler continually scans channels by monitoring quality of service reports on the signaling channel in the form of bit error rates, carrier to interference levels and signal to thermal noise level. Channels with high quality of service margins are available for use by bandwidth-efficient modems. When poor quality of service exists, modems are reassigned to clearer channels. More than likely, legacy modems will use the clearer channels. The bandwidth-efficient modems that have been added to the network can operate on channels where considerable interference exists, particularly from other bandwidth-efficient modems.

[0102] Table D is a table showing the channel assignment matrix. The modems are divided into two classes to address the situation where bandwidth-efficient modems have been added to a wireless network where other modems already operate without the ability to demodulate digital signals in the presence of strong interference 4

[0103] There are four cases corresponding to the two possible transmission and reception modems on the channel. Each modem has a common minimum signal to thermal noise requirement necessary for successful operation of the link. This SNR value is well known from digital communication theory and can be derived using Cramer-Rao bounds from statistical estimation theory. The value shown is representative for a particular modulation and encoding scheme. Actual values will vary accordingly. The minimum carrier-to-interference value is what differs between legacy modems and bandwidth-efficient modems. A legacy modem will treat interference as random thermal noise and be unable to demodulate its intended transmitter in the presence of interference comparable to maximum allowed thermal noise levels.

[0104] In contrast, a bandwidth-efficient modem is designed to jointly demodulate its intended transmission and any interfering transmission provided both signal levels are some minimum amount above the thermal noise power level. Therefore, a bandwidth efficient modem can operate in environments where its carrier to interference ratio is much less than one, although the actual value will depend on implementation details.

[0105] FIG. 6 shows an example where six satellite terminals 202 , 204 , 206 , 208 , 210 , and 212 share communications through a transponder 200 . There are three legacy terminals 202 , 204 , 206 and three bandwidth-efficient terminals 208 , 210 , and 212 . All communication is full duplex, so each link requires two channel assignments. The channels and terminal assignments are assigned according to the rules given in Table E. As a result, only four physical communication channels are necessary to provide 6 communication links. Also, given sufficient signal margin on channels 1 to 3, additional modems may be added. 5

[0106] Each legacy terminal gets a dedicated receiving channel. Due to their abilities to demodulate channels in the presence of interference, each applique modem 208 , 210 , 212 shares a channel. Given sufficient SNR margin on the legacy receive channels, further channels could be shared by having the applique modem signals buried at low power below the legacy channels. According to Table D, 18 dB SNR margin would be required to ensure that an applique modem at 9 dB SNR meet the 9 dB CIR requirement for the legacy modem and the 9 dB SNR requirement for the applique modem. Other modulation and encoding schemes would have different SNR and CIR threshold parameters.

[0107] Referring to FIGS. 7 a and 7 b, a wireless LAN network is depicted. While the satellite transponder generally transmits balanced power to the receiving terminals, the terminals on a wireless link do not receive equal power. There are essentially two basic types of LAN networks, a star network and mesh network. In the star network of FIG. 7 a, all communications go through a central hub 250 . A mesh network shown in FIG. 7 b, the various terminals 310 , 320 , 330 , 340 are interconnected with the controller 300 assigning the links. In either case, the hub must supervise and manage the communications process. For example, in a TDMA systems, if one terminal takes a time slot for communications, the hub should not otherwise schedule an overlap.

[0108] In the mesh network, the control element 300 is selected to avoid the ‘hidden terminal’ problem, wherein one of the terminals may be obscured from the field of view of one or more of the other terminals. This requirement is satisfied by periodically checking that the controlling element is able to monitor quality of service reports transmitted by all other terminals.

[0109] An advantage of the present system is that any bandwidth efficient terminal can function as the control element. If one of the bandwidth efficient terminals malfunctions or no longer has a good field of view, another terminal having better operating characteristics can assume control.

[0110] One of the hubs takes the primary role of managing, listening, and scheduling of the channels. For example, in a TDMA system, every frame consists of a control section and a data section. The present invention is primarily intended to improve data transfer, however inefficiency in the control section is tolerated to prevent communications from catastrophic failure. One terminal takes over as the network controller and sends the various sync signals, and processes book-keeping requests from the other terminals. In the mesh network, any terminal can be the network controller. In a hub network, all communication routes through a central network element, so access to all network elements is assured.

[0111] In satellite systems, the transponder transmits balanced and regulated power. Every signal looks the same when picked up by the various terminals. This provides a built in redundancy so that if one of the network controllers fails to function, an alternative terminal can take over and cause an automatic bootstrap.

[0112] In a wireless system, there is no singular point of balancing and regulating the transmission signals. The system can pick one LAN element as the controller and monitor signal quality. It should be readily apparent that it is difficult to measure the quality of service every terminal experiences. The processing of the present invention may require running the algorithm on every terminal that gas multi user capability and have it report back to the controller element, which evaluate the data from every terminal. Otherwise, the ability to schedule terminals on the same channels where excess SNR margin is enjoyed is the same as already described.

[0113] An additional application is in the cable modem field, wherein a cable network has excess linear power margin. The cable modems can interconnect to an individual house for example and provide the television and Internet delivery. Cable modems share channels and generally the cable is not saturated and there is sufficient room for adding more signals and still be in a linear state where the receiver unit is not saturated.

[0114] In a system wherein all the modems tolerate a level of interference, incorporating a network of modems with joint detection capability allows these modems to extract more signals such as illustrated in FIG. 6 . The cable node or head end would be the controller/scheduler for a plurality of cable modem. The available power for additional users is the difference in the power consumed by an existing transmission and the point where the power in the channel goes non-linear, thus defining the linear power limit. Channels can be added to the available bandwidth until the linear power region of the network is consumed. This results in increased channel throughput by adding multiple carriers to the same channel bandwidth. The additional capacity can be gained by using different coding schemes or modulation schemes to place the data on the channel and therefore extract the data from the channel by reversing the process thus increasing the effective throughput of the bandwidth consumed by the initial channel. For example, if the linear power limit was established, and the channel power margin consumed by a single transmission was known, then the available power for additional users is readily available.

[0115] The present invention is intended for a multiplicity of protocols involved in frequency reuse technology. The reference to multi user and multi user environments in the present application is not to be limited to referring only to multiple users. The term multi user is applied in the broader context of referring to any type of additional user or signals, whether individual users in a cell phone situation or additional information signals for a given number of users.

[0116] Numerous characteristics and advantages have been set forth in the foregoing description, together with details of structures and functions of the present invention, and some of the novel features thereof are pointed out in appended claims. The disclosure, however, is illustrative only, and changes may be made in arrangement and details, within the principle of the invention, to the full extent indicated by the broad general meaning of the terms in which the appended claims are expressed.

[0117] The objects and advantages of the invention may be further realized and attained by means of the instrumentalities and combinations particularly pointed out in the appended claims. Accordingly, the drawing and description are to be regarded as illustrative in nature, and not as restrictive.