System and Method for Sounding in a Full-Duplex CATV Architecture

Abstract

A system and method for managing interference groups in a full-duplex transmission network.

Description

Background Art

[0001] Cross - reference to Related Applications This application claims priority under 35 U.S.C. § 119(a) to U.S. Provisional Patent Application No. 62 / 911,081, filed Oct. 4, 2020, and U.S. Provisional Patent Application No. 62 / 937,913, filed Nov. 20, 2019, both of which are hereby incorporated by reference in their entirety.

[0002] The subject matter of this application relates to systems and methods for forming a group of cable modems into an interference group to facilitate full - duplex transmission in a CATV architecture.

[0003] Cable television (CATV) service provides content to a large group of subscribers from a central distribution unit, called a “head - end”, which distributes channels of content to subscribers via a branched network that includes a number of intermediate nodes. However, modern cable television (CATV) service networks not only provide media content such as television channels and music channels to customers, but also host digital communication services such as Internet service, video - on - demand, and telephone services such as VoIP. These digital communication services require not only downstream communication from the head - end, through intermediate nodes, to subscribers, but also upstream communication from subscribers, through the branched network, to content providers.

[0004] For this purpose, a CATV headend has conventionally included a separate cable modem termination system (CMTS) used to provide high-speed data services such as video, cable Internet, voice over Internet protocol, etc. to cable subscribers. Typically, the CMTS includes both an Ethernet (registered trademark) interface (or other more conventional high-speed data interfaces) as well as an RF interface, such that traffic coming from the Internet can be routed (or bridged) through the Ethernet (registered trademark) interface, through the CMTS, and then to an optical RF interface connected to the cable company's hybrid fiber coaxial (HFC) system. Downstream traffic is sent from the CMTS to the cable modem at the subscriber's home, while upstream traffic is sent from the cable modem at the subscriber's home to the CMTS. Many of the latest CATV systems combine the functions of the CMTS and the video distribution system (EdgeQAM) into a single platform called a converged cable access platform (CCAP). Additionally, other latest CATV systems called remote PHY (or R-PHY) relocate the physical layer (PHY) of a conventional CCAP by pushing it to the fiber nodes of the network. Thus, while the CCAP core performs upper layer processing, the R-PHY devices within the node convert the downstream data transmitted by the core from digital to analog for transmission at radio frequencies, and convert the upstream RF data transmitted by the cable modem from analog to digital for transmission to the core over optical.

[0005] Regardless of whether such an architecture has been adopted, past implementations of CATV systems have split the available bandwidth between upstream and downstream transmissions, i.e., data has been transmitted in only one direction across any portion of the spectrum. For example, initial iterations of the Data Over Cable Service Interface Specification (DOCSIS) allocated upstream transmission to the frequency spectrum from 5 MHz to 42 MHz and downstream transmission to the frequency spectrum from 50 MHz to 750 MHz. In later iterations of the DOCSIS standard, the width of the reserved spectrum for each of the upstream and downstream transmission paths was extended, but the spectra allocated to each respective direction did not overlap.

[0006] Recently, cable operators have been looking for alternative architectures to provide multi-gigabit services. This need, along with recent trends in the cable industry such as the deployment of DOCSIS 3.1 Orthogonal Frequency Division Multiplexing (OFDM), deep fiber migration, and Remote PHY network architectures, has led to the development and standardization of Full Duplex (FDX) DOCSIS technology. In FDX DOCSIS, there is no need to separate the upstream and downstream spectra, enabling upstream services of up to 5 Gbps and downstream services of up to 10 Gbps on the cable access network. In a full duplex system, the CCAP / R-PHY core knows the characteristics of its own downstream transmissions and can thus distinguish upstream communications transmitted at the same frequency as the downstream services it provides.

[0007] However, in an FDX system, interference between bi-directional transmissions must be reduced so that the intended downstream signals are properly received. In a point-to-multipoint system where multiple cable modems (CMs) are connected to the same cable modem termination system (CMTS) port, when one CM transmits upstream to the CMTS, the upstream signal may leak through the cable plant and interfere with the reception of downstream signals received by other cable modems. Since the source of interference is unknown to the receiving cable modem, techniques such as PHY layer echo cancellation cannot be used.

[0008] Accordingly, what is desired is an improved system and method for reducing interference in a full-duplex CATV transmission architecture. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] For a better understanding of the present invention and to show how the same may be carried out, reference is now made, by way of example, to the following accompanying drawings.

Figure 1

Figure 2A

Figure 2B

Figure 2C

Figure 2D

Figure 3

Figure 4A

Figure 4B

Figure 5

Figure 6A

Figure 6B

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

[0010] In the first embodiment, the CATV system may include a headend connected to a plurality of modems via a transmission network, and the plurality of modems are arranged within a first set of at least one interference group (IG). The system may include a processor that is operably connectable to the headend and that can reconfigure the plurality of modems within a second set of at least one IG that is different from the first set of at least one IG. The processor may reconfigure the plurality of modems based on the first set of at least one IG.

[0011] In some embodiments of the foregoing CATV system, graph-based connection component technology may be used to process an array based on sounding data to reconfigure a plurality of modems into a second set of at least one interference group. The graph may be undirected or directed. In some situations, graph-based connection component technology may be applied only to a subset of the sounding data, and the subset is based on identified incremental changes from the first set of at least one interference group. In some embodiments, the identified incremental changes may include pairs of modems in different IGs that currently interfere with each other and did not interfere with each other at a previous time. In some embodiments, the identified incremental changes include pairs of modems within the same IG that no longer interfere with each other.

[0012] The graph-based connection component technology may be based on a graph having nodes and edges. The nodes may correspond to individual cable modems, and the edges may connect the nodes, each having an associated interference measurement criterion measured by sounding procedures on the CATV system.

[0013] In a second embodiment, the method may place a plurality of modems in a CATV system. The method may automatically place a plurality of modems in a first placement of at least one interference group by a processor operably connected to a head end of the CATV system. The processor may automatically reconfigure the plurality of modems to a second placement of at least one IG that is different from the first placement of the at least one IG using the first placement.

[0014] In some embodiments of the CATV system described above, a graph-based connection component technique may be used to process an array based on sounding data to reconfigure a plurality of modems to a second placement. The graph may be undirected or directed. In some situations, the graph-based connection component technique may be applied only to a subset of the sounding data, and the subset is based on identified incremental changes from the first placement. In some embodiments, the identified incremental changes may include pairs of modems in different IGs that currently interfere with each other and did not interfere with each other at a previous time. In some embodiments, the identified incremental changes include pairs of modems in the same IG that no longer interfere with each other.

[0015] Some embodiments may use a graph having nodes and edges. The nodes may correspond to individual cable modems, and the edges may connect the nodes, each having an associated interference measurement criterion measured by a sounding procedure on the CATV system.

[0016] In the third embodiment, the processing system can exchange data between the head end and a plurality of cable modems via a transmission network. The processing system includes a processor that selectively initiates a periodic baseline sounding test of the cable modems and a periodic full sounding test of the cable modems, where the full sounding test generates more data than the baseline sounding test. The processing system may include a memory that stores at least one past baseline sounding test. The processor can perform a current baseline sounding test and use a comparison between the current baseline sounding test and at least one record of past baseline tests to select whether to initiate a full sounding test.

[0017] The above-described processing system can perform a baseline sounding test that measures the noise at each cable modem when the cable modem is not transmitting data. The baseline sounding test can generate a record that includes a vector of noise measurements for each band in a plurality of frequency bands across the spectrum of interest.

[0018] The processing system may include a database that stores a plurality of past baseline sounding records. The processor can use one selected record from the plurality of past baseline sounding records to compare with the current baseline sounding test and use the comparison to select whether to initiate a full sounding test. For example, at least one record of a past baseline test can include at least one of the days of the week for each at least one record and the time of each at least one record. Alternatively or additionally, at least one record of a past baseline test can include at least one meteorological measurement criterion associated with each at least one record, such as ambient temperature data that occurred during the baseline sounding test associated with the record. In some embodiments, such measurement criteria can be obtained from a meteorological database after at least one record of a past baseline test has been created.

[0019] The fourth embodiment may include a method for exchanging data between a head end and a plurality of cable modems via a transmission network. The method may selectively initiate periodic baseline sounding tests of the cable modems and periodic full sounding tests of the cable modems, and the full sounding tests generate more data than the baseline sounding tests. At least one record of past baseline sounding tests is stored in memory, and then the current baseline sounding test may be performed. A comparison between the current baseline sounding test and at least one record of past baseline tests may be used to select whether to initiate a full sounding test.

[0020] The method of the fourth embodiment may perform a baseline sounding test that measures noise at each cable modem when the cable modem is not transmitting data. The baseline sounding test may use a record that includes a vector of noise measurements for each band in a plurality of frequency bands across the spectrum of interest.

[0021] The method of the fourth embodiment uses a database that stores a plurality of past baseline sounding records, uses one selected record from the plurality of past baseline sounding records for comparison with the current baseline sounding test, and uses the comparison to select whether to initiate a full sounding test. For example, at least one record of past baseline tests may include at least one of the days of the week for each at least one record and the time of each at least one record. Alternatively or additionally, at least one record of past baseline tests may include at least one meteorological measurement criterion associated with each at least one record, such as ambient temperature data that occurred during the baseline sounding test associated with the record. In some embodiments, such measurement criteria may be obtained from a meteorological database after at least one record of past baseline tests has been created.

Best Mode for Carrying Out the Invention

[0022] As previously described, the DOCSIS specification has used different frequency bands for upstream and downstream data traffic. Even when multiple cable modems in a particular service group share the same network resources, upstream and downstream traffic is completely separated. Recently, as an attempt to provide symmetric services in both upstream and downstream, a new FDX (Full Duplex) standard has been introduced to simultaneously use a portion of the coaxial network bandwidth for both upstream and downstream traffic. In the FDX architecture, the CMTS can receive and transmit simultaneously within the same FDX spectrum, while an FDX cable modem can perform either reception or transmission within the same FDX spectrum, but not both. The FDX band is divided into sub-bands, and the CMTS assigns the sub-band(s) that each cable modem uses for upstream or downstream operations. This is referred to as Resource Block Assignment (RBA). Different cable modems have different bandwidth requirements for both upstream and downstream directions, which may change over time, and FDX can dynamically change the RBA accordingly. Therefore, the communication is full duplex from the perspective of the CMTS, but frequency division duplex from the perspective of the cable modem.

[0023] However, in the FDX architecture, RF signals from a modem transmitting data upstream can interfere with other modems receiving data downstream. Such interference can be minimized by configuring the modems into interference groups. Referring to FIG. 1, for example, a CATV transmission architecture 1 may include a CCAP 2 at a head end connected to a plurality of cable modems 4 via a branch transmission network. The architecture of FIG. 1 is shown as an R-PHY system where the CMTS operates as a CCAP core while the remote physical device (RPD) is located downstream, but alternative systems may use a conventional CCAP that operates entirely with an integrated CMTS at the head end connected to the cable modems 4 via a plurality of nodes / amplifiers.

[0024] Preferably, to facilitate FDX transmission, cable modems are configured into interference groups (IGs) 5, 6, 7, 8, etc. As the name indicates, an IG is a set or group of modems where the upstream transmission of one or more of the modems within the IG interferes to an unacceptable degree with the downstream reception of other modems within the IG, but does not interfere to an unacceptable degree with the downstream transmission of cable modems within other IGs. Identifying these IGs and using the IG groups to schedule downstream and upstream transmissions appropriately is essential for achieving the high throughput of the FDX system by enabling the CCAP to schedule downstream transmissions to all cable modems within an IG when the cable modems within the IG are not transmitting upstream.

[0025] To facilitate the configuration of cable modems into IGs, sounding techniques can be used to measure the interference caused to other cable modems in the network by the upstream transmission of a particular cable modem. During sounding, a given modem transmits a pilot signal upstream while the remaining modems in the service group measure the downstream modulation error rate (RxMER). This process is repeated by different transmitting modems, resulting in a matrix indicating the interference on the same channel for the entire service group.

[0026] In some embodiments, sounding data may be collected from a plurality of CCAP cores from a number of service groups, and the collected data may be processed by the centralized processor 9 shown in FIG. 1 to configure the cable modems into respective IGs. Therefore, a very efficient algorithm that is scalable with a very large number of SGs is required. Further, data may be repeatedly collected from the same set of SGs. For this, an algorithm that can efficiently process incremental data is necessary.

[0027] A given service group (SG) served by a CMTS / CCAP system may have dozens or hundreds of cable modems. FIGS. 2A-2C show a novel graph-based approach for efficiently configuring such a large number of cable modems into IGs using sounding data. In this approach, a network graph is created in which individual modems form the nodes of the graph. The edges of the graph are the RxMER metrics obtained as a result of the sounding process.

[0028] As an example, FIG. 2A shows a graph 10 including ten nodes 12, each corresponding to an individual modem in a network served by a CMTS / CCAP. Each cable modem is connected to other cable modems in the network by edges 14, which quantitatively show the interference between two connected modems as measured in the sounding procedure. In reality, each node 12 in FIG. 1 should be connected to all other nodes 12 by edges 14 because transmission by any cable modem causes at least some noise to all other cable modems. For simplicity, FIG. 1 omits those edges 14 that are so low that the MER measured by the sounding procedure is presumed not to cause interference.

[0029] For simplicity, FIG. 2A further shows an undirected graph where the edges 14 between nodes 12 have no direction. Another more realistic embodiment is a directed graph 16 as shown in FIG. 2B, where the edges 14 have a direction 15 indicated by an arrow. An edge 14 in the direction from node A to node B represents the interference observed by node B when node A is transmitting a beacon signal during the sounding algorithm, while an edge 14 in the direction from node B to node A represents the interference observed by node A when node B is transmitting a beacon signal during the sounding algorithm. One skilled in the art will understand that the weight of the edge (interference) in the direction from A to B need not be the same as the weight of the edge (interference) in the direction from B to A. One skilled in the art will also understand that in some cases, there may be an edge only in one direction between two nodes and not in the other direction (e.g., between nodes G and C, it is shown that there is interference observable at node G when node C is transmitting, while there is no interference observable at node C when node G is transmitting). Such an asymmetry may be less likely but is still possible and can be incorporated into the disclosed embodiments.

[0030] A directed graph can be reduced to an undirected graph for simplification of calculations based on various factors. For example, a directed edge between nodes A and B can be replaced by an undirected edge having a weight value equal to the average value, maximum value, or weight value of the two directed edges between nodes A and B, or any other appropriate measure. For example, the undirected graph of FIG. 2A utilizes the maximum value of the directional weights of FIG. 2B.

[0031] Analytically, the disclosed graph-based solution consists of two steps. The first step is to create a graph as described above based on data collected during the sounding process, optionally by using a first directed graph to generate an undirected graph. The second step is to derive interference groups by using a connected component algorithm on the graph to compose the original graph into one or more subgraphs (SGs), such that in each individual SG, (i) there exists a path between any pair of nodes within the SG, and (ii) there is no path between two nodes in two different SGs. Next, the SGs can be used as IGs. The connected component algorithm can be applied to both directed and undirected graphs. For simplicity, the remainder of this disclosure provides an exemplary connected component analysis on an undirected graph. However, one skilled in the art will recognize how to modify such a method to be used on a directed graph.

[0032] For example, the connected component algorithm applied to the undirected graph of FIG. 2A generates a single IG that includes all modems because there exists a path connecting each node to every other node. However, here, assume that based on the modulation scheme used to transmit data between the cable modems represented by the nodes, the modems can withstand an RxMER level of 25 dB or less. Under this assumption, the link between node A and node G can be removed. Applying the connected component algorithm to this modified graph results in two SGs 18 and 20 as shown in FIG. 2C, and the SGs will correspond to the IGs defined by the FDX standard. Different levels of complexity of the connected component algorithm using a depth-first or breadth-first search can be applied to obtain the IGs shown in FIG. 2C.

[0033] Figure 2D shows how this graph-based connection component algorithm can be directly implemented on the sounding interference matrix S. Specifically, the sounding matrix S shown in Figure 2D corresponds to the data shown in Figure 2A. The slashes mean that the individual modems do not interfere with each other. Starting arbitrarily from node A and proceeding through the matrix, for example, node A is first added to subgraph IG1, and then subgraph IG1 is completed by iteratively tracing all paths starting from each node previously added to the subgraph until the path to another node already in the subgraph ends or loops. Thus, matrix X shows that node A is connected to node B, node B is connected to node D, node D is connected to node E, and node E is not connected to any further nodes. Going back through the path, neither node D nor node B is connected to any additional nodes not yet in matrix S. Going further back, node A is connected to node D which is already included, but is also connected to node G, and node G is added to subgraph SG1. Continuing this procedure from node G, nodes C, F, I, J, and H are successively added to subgraph IG1 until all nodes are in IG1, thus replicating the graph shown in Figure 2A.

[0034] However, again here, RxMER measurement values of 25 dB or less are acceptable, and the values of "25" and "22" as shown in Figure 2B corresponding to the interference between nodes A and G are replaced with zero, and then assuming that two IGs, IG1 and IG2, have been generated by the procedure described above. Specifically, as described above, tracing the path through matrix S from node A would successively add nodes A, B, D, and E to subgraph IG1, but then no further nodes are added because there is no longer an MER value to connect node A to node G. Thereafter, a new subgraph SG2 is created, and a path starting from the next node C is traced that successively adds nodes G, F, I, J, and H to subgraph SG2, thereby generating the result shown in Figure 2C.

[0035] Referring to FIG. 2E, the foregoing procedure can be summarized by a process 30 that collects sounding data at step 32. In an optional step 34, a threshold value is applied to the entries of the sounding data such that it reflects the boundary between acceptable and unacceptable interference between cable modems. At step 36, a sub-graph (i.e., IG) is created by iteratively tracing a path through non-zero values within the sounding data. One of ordinary skill in the art will understand that while one such iterative path tracing technique is described in the foregoing paragraph, other such iterative techniques can be readily substituted. As an example, instead of tracing one complete path at a time and then moving to another node within the sub-graph to completely trace that path to fill the sub-graph, other techniques trace all paths starting from the first node added to the sub-graph in parallel. For example, from the initial node A of the matrix S in FIG. 2D, all nodes B, E, and G are added to a subgroup (from node A), and then nodes D (traced from either node B or node E), and nodes C, F, and H (traced from node G), etc. can be added. Further alternatively, the matrix S may simply reflect 1s and 0s to indicate acceptable or unacceptable interference, or the threshold value may be applied to each step of the path construction process through the matrix S rather than being used in a separate step to change the matrix S.

[0036] Each time data from a sounding procedure is received, a connection component algorithm can be launched to identify IGs. However, since data is collected periodically, the required computational effort increases significantly, and a brute force application of the connection component algorithm does not take advantage of the fact that most IGs are likely to remain unchanged in the execution of successive data collections. Taking advantage of this, some embodiments of the present disclosure may use any of a number of incremental processing techniques. To facilitate such techniques, the sounding data from the most recent past sounding procedure, and any IGs and / or other results obtained from the connection component algorithm obtained from the most recent past procedure are each selectively stored and then can be used in a next iteration to reduce the computational requirements.

[0037] Let G be a graph that includes all nodes within the FDX group. If N is a list of all nodes (modems) within the FDX group and E is a list of all edges indicating interference between modems, G = (N, E).

[0038] When the connection component algorithm is executed, the graph G is subdivided into a plurality of connection components (IGs) or subgraphs G1, G2,... G k Thus, G = G1 U G2 U... U G k (where U is the union operation).

[0039] The complexity of the connection component algorithm scales with the number of nodes and edges within a given graph, i.e., the complexity of the algorithm is proportional to O(|N| + |E|), where |N| is the number of nodes within the list N and |E| is the number of edges connecting any pair of nodes. Thus, executing the connection component algorithm after every iteration of data collection is preferably avoided, or alternatively, the connection component algorithm is executed on a smaller subset of the data to reduce the computational complexity.

[0040] To simplify the following discussion, assume that the binary matrix M(n) represents interference information based on data collected at iteration n. The matrix M(n) is of size |N|x|N|, and the entry in the i-th row and j-th column of the matrix M(n) is 1, indicating that Modem i interferes with Modem j and 0. Figure 3 shows the matrix M(N) corresponding to Figure 2A.

[0041] A simple technique to minimize computational requirements is to store the result of M(t - 1) and then compare M(t) with M(t - 1). If there is no change between M(t) and M(t - 1), then there is no need to execute the connected component algorithm on the sounding data collected during iteration t. This simple check may prove true for multiple iterations, resulting in no additional computation required. However, there are more subtle and smaller changes in the interference data that can be used to further reduce computational requirements.

[0042] Specifically, there are two types of incremental changes that can be made to a series of subgraphs created by the connected component algorithm of the present disclosure. The first incremental change (type 1) is when a modem in IG group G i may start interfering with a modem in another group G j at iteration t. Due to this interference, these two groups G i and G j can be merged to create a larger IG. One skilled in the art will understand that if a modem starts interfering with another modem within the same IG that did not interfere at the previous iteration t - 1 (and nothing else has changed), this will not cause any change to that IG at iteration t.

[0043] The second incremental change (type 2) occurs when a modem in IG group G i stops interfering with another modem within the same IG. This causes the IG group G ican be divided into two smaller IG groups. Those skilled in the art will understand that in some embodiments, type 2 changes may be ignored or selectively corrected. However, type 1 changes cannot be ignored because they interfere with the throughput performance of FDX.

[0044] The correction of type 2 incremental changes is considered first because it is easier to analyze. Referring to FIG. 4, the interference matrix I(IG for individual subgraphs or IGs i) can be constructed from sounding data stored stepwise before the sounding of the current round. This matrix identifies modems / nodes that interfere with each other as "1" and modems / nodes that do not interfere with each other as "0". Thus, the matrix shown in FIG. 4 reflects the IG 18 shown in FIG. 2C, where each row and column corresponds to nodes A, B, D, and E of matrix M in FIG. 3. The second matrix is constructed for that IG using the sounding of the current round, except that in this second matrix, modems that interfere with each other are represented as "0" and modems that do not interfere with each other are represented as "1". Thus, by performing a logical "AND" operation on these two matrices to generate a third matrix, any value "1" in the matrix corresponds to a pair of modems that began interfering with each other at time t-1 in the same IG but no longer interfere with each other at time t. By continuously applying this procedure to each individual SG and adding an array T2 (also shown in FIG. 4) using any IG in which a "1" appears in the matrix resulting from the AND operation, the array T2 enumerates the SGs that are candidates for further division. Next, the connected component algorithm can be applied individually only to the smaller IGs identified in the array T2. One skilled in the art will understand that individual subgraphs (IGs) tend to be much smaller than the original graph G and thus result in computational savings. Further, as already mentioned, in some embodiments, detecting and optimizing type 2 changes need not be performed for every sounding algorithm iteration. One skilled in the art will understand that the foregoing procedure can be easily modified to construct the second matrix exactly as the first matrix in FIG. 4 was constructed by alternative procedures, simply by comparing the two ways and identifying any changes regardless of whether the pair of modems switched from interference to non-interference or vice versa.

[0045] Figure 5 shows a method 40 for implementing the procedure described above. In step 42, a null array T2 is initialized that contains the value "0" for each IG identified in a previous implementation of the connection component algorithm. In step 44, a first matrix, i.e., an interference matrix as described above, is constructed for the first IG from data obtained in the previous round of sounding. In step 46, a second matrix, i.e., a non-interference matrix, is constructed for the first IG as described above from data obtained in the current round of sounding. In step 48, a logical AND operation is performed on these two matrices to generate a third matrix, or otherwise the two matrices are compared. From the comparison in step 48, in step 49, it is determined whether a necessary change has occurred between the first matrix and the second matrix, e.g., whether two modems that were interfering with each other are no longer interfering, or whether some change in the interference state has occurred from one round of sounding to the next. If the answer is "yes", the value of that IG in array T2 is changed and the procedure proceeds to the next IG in the array. If the answer is "no", the procedure proceeds to the next IG in the array without changing the value of array T2. If the IG does not change, array T2 may be used to identify the IGs that can be re-executed through the connection component algorithm again.

[0046] Referring to FIGS. 6A and 6B, different procedures can be used to identify Type 1 changes. Specifically, referring to FIG. 6A, the interference group matrix I can be constructed from the previous round of sounding data that indicates, for all modems or nodes within the collection of IGs, whether a particular modem / node is in the same IG as another modem / node. That is, if modem i (row i of the matrix) and modem j (column j of the matrix) are in the same IG, the corresponding entry is 0, and 1 otherwise. Since this data exists from the previous round of sounding, little additional processing is needed to construct this matrix. As can be seen, matrix I of FIG. 6A corresponds to the subgraph shown in FIG. 2C, which was constructed using matrix M of FIG. 3.

[0047] Next, a second matrix M as shown in FIG. 3 is constructed using sounding data from the current round of sounding, and a logical “AND” operation is applied to the two matrices, thereby identifying any instances where the two cable modems are interfering with each other and are not currently within the same IG. After performing the logical AND operation, if any of the entries in row i (or column j) is 1, the IG to which node i (or j) belonged in the previous iteration needs to be added to array T1 (shown in FIG. 6B). Once array T1 is fully added, the connected component algorithm can be applied to the union of the IGs identified in array T1. One of ordinary skill in the art will understand that instead of first constructing matrix I as seen in FIG. 6A, the individual arrays associated with each node / modem can be constructed to identify (1) other nodes / modems within the same SG as that modem, and (2) other nodes / modems that are currently interfering with that individual modem. Next, an “AND” operation can be applied to these arrays, add the associated SGs to array T1, and move the procedure to the next node / modem, etc. until array T1 is fully added. One of ordinary skill in the art will also understand that regardless of the technique we used, this graph is substantially smaller than the original graph G due to the slight changes in sounding interference in successive iterations and the fact that the complexity of the connected component algorithm is proportional to O(|N|+|E|), and thus will require fewer computational steps in most cases.

[0048] FIG. 7 shows a method 60 for implementing the procedure just described. At step 62, a null array T1 is initialized that contains the value “0” for each IG identified in a previous implementation of the connection component algorithm. At step 64, an I matrix indicating which node / modem pairs are in different IGs, as described above, is constructed from data obtained in the previous round of sounding, and at step 66, a matrix M (as disclosed in FIG. 3) is constructed as described above from data obtained in the current round of sounding. At step 68, a logical AND operation is performed on these two matrices to produce a third matrix whose value “1” indicates node / modem pairs of different IGs that are currently interfering with each other. At step 60, this third matrix is used to add to array T1 and identify any IG having a node / modem flagged in the third matrix. At step 62, the connection component algorithm is applied to the reconciliation of the IGs in T1.

[0049] One of ordinary skill in the art will understand that the procedures just described can be combined to still further provide additional efficiency. For example, if the procedure used to identify type 2 changes identifies only instances where the modem / node pairs within an IG have switched from interfering with each other to not interfering with each other, the connection component algorithm can be applied to potentially split those identified SGs. Then, type 1 changes can be identified by constructing the I matrix shown in FIG. 6A using the new IG organization. Then, since it is known that each IG cannot be further subdivided, when the type 1 procedure then proceeds to add to the T1 array, the values within the array can be used to directly merge the IGs without the need for the connection component algorithm.

[0050] Alternatively, a change of type 1 can be identified by first adding to the T1 array and then identifying to execute a connection component algorithm on the union of the IGs identified in the T1 array. Since the connection component algorithm itself constructs all the modems / nodes within these IGs into the smallest possible combinations, the procedure for a change of type 2 needs to be executed only on the remaining IGs not included in the T1 array.

[0051] The foregoing systems and methods correct the inefficiencies in processing sounding data to dynamically assign cable modems to one or more interference groups (IGs), but the sounding process itself imposes a significant overhead on the transmission system. As described above, in the sounding procedure, the CMTS instructs one or more FDX-capable CMs to transmit a test signal on a designated subcarrier, while instructing other FDX-capable CMs to measure and report the received RxMER on the same set of subcarriers. The CMTS repeats this procedure using other CMs as transmitters until the interference levels are tested between all combinations of CMs. Further, the CMTS can repeat this over all relevant subcarriers.

[0052] In the sounding commonly used in the FDX system, there are two types of sounding: continuous wave (CW) sounding and OFDMA upstream data profile (OUDP) sounding. During CW sounding, one or more test cable modems transmit CW test signals at selected subcarrier frequency positions (each cable modem supports up to 255 subcarriers), while the remaining cable modems measure the RxMER of the zero-bit load downstream signal received simultaneously with the upstream test transmission. These measurements ideally include up to 3800 subcarriers, including the subcarriers of the test CW signal. The advantage of CW sounding is that interference at these frequencies is tested independently, so a relatively small number of subcarrier frequencies are combined at once. This allows the remaining part of the subcarriers to be used for content delivery. The disadvantage of CW sounding is the length of time it takes to complete the procedure, which can take up to several minutes. During this time, full use of the available spectrum is excluded.

[0053] Conversely, OUDP sounding occupies the entire spectrum for each test burst from each cable modem, where the test burst may last about 20 - 60 milliseconds, and each test burst includes 3800 measurements, one for each 50KHz band in the spectrum. Even when repeated for a large number of modems, the overall procedure is much faster than CW sounding. However, in this procedure, the appropriate OFDM spectrum of the CATV plant cannot be used during the OUDP test burst because the burst covers the entire OFDM channel. Regardless of whether CW or OUDP sounding is used, customers may experience jitter or a reduction in bandwidth during peak hours of the day due to the sounding round.

[0054] A novel system and method are disclosed for reducing the frequencies at which any of the foregoing types of sounding are required. This procedure, herein, performs what is referred to as a "baseline" sounding test in which all cable modems measure the noise level that exists when signals are not being transmitted in the upstream or downstream direction. In this procedure, for each cable modem, noise measurements at each subcarrier frequency used by the cable modem are collected in a "baseline CW" test while signals are not being transmitted by the cable modem. The "baseline OUDP" test collects 3800 measurements in 50KHz increments across the entire spectrum utilized by the system for each cable modem. Compared to full sounding procedures such as CW and OUDP sounding, baseline sounding requires far fewer system resources. Baseline sounding essentially measures the noise floor of the transmission path between the headend and the customer's cable modem caused by factors such as standing wave reflections along the transmission path and spurious electromagnetic interference that varies based on the length of the transmission path, e.g., ambient weather conditions such as temperature. However, many of these factors do not change over time, such as the transmission length between the headend and a given cable modem, and when changes occur during continuous baseline sounding measurements, these changes often strongly correlate with changes in the network topology, such as when the customer's cable modem comes or goes offline, which is often the source of changes during full sounding measurements. Accordingly, the inventors recognized that instead of simply using the periodic sequence of full sounding tests, the disclosed baseline sounding procedure can be used to determine whether full sounding is necessary.

[0055] Referring specifically to FIG. 8, a system such as that disclosed in FIG. 1 may use a method 70 that performs a baseline sounding test and stores the results at step 72. The baseline sounding test covers all 3800 frequencies of the OFDM / OFDMA band, while no transmission occurs. At step 74, a complete sounding test is performed and the results are stored. At step 75, the complete sounding results are used to create a set of interference groups (IGs). At step 76, a baseline sounding test is performed after an appropriate interval such as hourly. At step 78, it is determined whether a sufficient change in the baseline sounding results has occurred to warrant a full round of sounding, compared to the results measured at step 72. For example, the installation of a new cell tower, recently exposed wiring (functioning as an antenna), or signals from a nearby HAM radio operator may each affect the baseline sounding measurements and the complete sounding measurements, and by detecting a change in the baseline sounding results, a complete sounding procedure can be initiated. Alternatively, a new modem may go online or offline, which requires a full round of sounding to determine at least which IG to add the new modem to or whether the removal of a modem guarantees the splitting of an IG. In any of these situations, the results measured by the baseline sounding procedure may include significant changes such as a change in the noise registered by any individual modem within a service group that exceeds a set threshold, indicating that the state of the plant has changed and warranting a new complete sounding test again. In any such situation where it is determined by comparison that a new round of complete sounding is warranted, at step 79, a new baseline result is stored and the procedure returns to step 74, where a complete sounding is performed, etc.

[0056] Conversely, if no significant change is observed between the baseline sounding result obtained in step 76 and the previous baseline sounding result, no change to the IG is required, and the procedure returns to step 76, where another baseline sounding is performed at the next scheduled interval. In some embodiments, if a transmission error is reported, a full sounding round may be triggered regardless of the lack of a significant change in the baseline sounding result.

[0057] As described above, the baseline sounding result may change over time due to factors not related to changes in system configuration (such as addition or removal of modems, system maintenance of part of the transmission network, etc.) or other such changes that may require a change in the interference group. As an example, the baseline sounding result may change based on time, ambient weather conditions such as temperature, and other similar factors. Thus, some embodiments of the present disclosure may store multiple different past baseline sounding results, such as different days of the week, different times of day, different temperatures, and other weather conditions. When a new baseline sounding round is performed, the most relevant one of the stored past results may be retrieved for comparison.

[0058] For example, referring to FIG. 9, such an embodiment may include a system 80 that includes a CCAP core 82 connected to a plurality of cable modems 86 within a customer's premises via the network of the RPD 84, where the cable modems 86 are assigned to interference groups 87 (only one of which is shown in FIG. 9). Similar to FIG. 1, the architecture of FIG. 9 is shown as an R-PHY system where the CMTS operates as the CCAP core while the remote physical device (RPD) is located downstream, but alternative systems may use a conventional CCAP that operates fully at the headend CMTS connected to the cable modem 4 via a plurality of nodes / amplifiers.

[0059] Preferably, the CCAP core 82, or other head-end device, may be connected to a database 88 that selectively stores past sounding data in a memory 89. FIG. 10 shows an exemplary scheme in which past baseline sounding records can be stored. Specifically, the baseline sounding data for each round is tagged with metadata indicating some or all of the following: date (which may include day of the week), time at which the baseline sounding data was collected, temperature data corresponding to the location of each of the CCAP, RPD, and cable modem, and other factors that may be considered relevant to determining which stored past baseline sounding record most closely represents the conditions for comparing the current baseline sounding data. For example, if the current baseline sounding test is performed at 5:00 PM, the system 80 can retrieve the past record taken at 5:00 PM the previous day. In other embodiments, as more data is collected, if the current baseline sounding test is performed at 5:00 PM on a Wednesday, the system 80 can retrieve the past record taken at 5:00 PM the previous Wednesday. As even more past data is collected, the system 80 may be able to add more filters, such as temperature and other weather conditions, or other data, so that past baseline sounding results can be obtained that correspond to the same time and most closely match the current weather conditions. In some embodiments, information such as weather conditions at a given day and time can be accessed from a weather server or other such available database and added to the baseline sounding record by using the retrieved data, and can be retrieved after the baseline sounding is performed.

[0060] In a preferred embodiment, the database 88 having past sounding data 89 may be remotely connected to the CCAP core 82, but in other embodiments, the CCAP core 82 may be integrated into the database 88. Similarly, some embodiments may include a management or processing function having a database 88 remotely connected to the CCAP core 82 such that the CCAP core 82 simply initiates a request for past sounding records for comparison with current results, and the manager / database 88 determines the most relevant records and returns the results to the CCAP core 82.

[0061] Figure 11 shows an exemplary procedure 90 used by the system of Figure 9. At step 91, a baseline sounding is performed, and at step 97, the results are stored in the database. At step 92, a full sounding round is performed, and at step 93, a series of interference groups are created based on the results of the full sounding round. At step 94, another baseline sounding round is performed, and at step 98, the results are stored in the database. At step 95, a request for past baseline records from the database is made, and at step 99, the results are identified and returned from the database. At step 96, the past records returned at step 99 are compared with the current baseline sounding results obtained at step 94. If the comparison shows that another round of full sounding is warranted, the procedure returns to step 92; otherwise, the procedure returns to step 94.

[0062] In some embodiments, the systems and methods shown in FIGS. 9 and 11 can employ a statistical model to make a determination as to whether the difference between the current round of sounding and the past baseline record can guarantee a full sounding for the new round. Specifically, when the systems and methods shown are first implemented, the database 88 shown in FIG. 9 can collect information about the baseline sounding results, the full sounding results, and whether the full sounding for the new round will cause a change in the interference group while the full round of sounding is repeatedly collected. Then, the system can begin to correlate the change in the measurement criteria between the current baseline sounding reading and the most relevant past record selected by the system (e.g., the maximum difference in the baseline sounding readings for any modem), and the possibility that the change was made in the IG. As more data is collected, the statistical model becomes more reliable, and when a certain level of reliability is reached, the system can begin to use the baseline sounding reading as a proxy for the full sounding round after the desired threshold probability (e.g., 90%, 95%, or any other desired threshold) that another full sounding procedure will cause a change in the interference group is calculated.

[0063] In other embodiments, the system operator can empirically determine what qualitative and quantitative changes in the baseline sounding readings are most likely to result in the need for a full sounding for the new round.

[0064] In some embodiments of the disclosed system, database 88 may store IGs associated with complete sounding results and baseline sounding results, and may select a new IG based on records in the past database without performing a complete sounding test. For example, if there is insufficient bandwidth to perform a complete sounding test, a baseline sounding test may be used as a proxy to temporarily select an IG. Alternatively, for example, if the CATV system has to be re-initialized after a software update, the past database may be queried for the latest set of IGs.

[0065] The present invention is not limited to the specific embodiments described, and variations thereof may be made without departing from the scope of the invention as defined in the appended claims, and may be carried out in accordance with the principles of superior law, including some other principles that extend the enforceable scope of the claims beyond the literal scope or equivalent doctrine. It will be understood that references in the claims to the number of instances of an element, unless the context indicates otherwise, require at least a predetermined number of instances of the element, whether it be a reference to one instance or a reference to a plurality of instances, and are not intended to exclude from the scope of the structure or method claimed having more instances of that element than are described. When used in the claims, the term "comprising" or its derivatives is used in a non-exclusive sense, not intending to exclude the presence of other elements or steps in the claimed structure or method. Some aspects of the present invention are described below. [Aspect 1] A CATV system including a headend connected to a plurality of modems via a transmission network, wherein the plurality of modems are arranged within a first set of at least one interference group (IG), the system being operably connectable to the headend, and including a processor capable of reconfiguring the plurality of modems within a second set of at least one IG different from the first set of at least one IG, the processor reconfiguring the plurality of modems based on the first set of at least one IG. [Aspect 2] The CATV system according to aspect 1, wherein the system uses graph-based connection component technology to process an array based on sounding data to reconfigure the plurality of modems into the second set of at least one interference group. [Aspect 3] The CATV system according to aspect 2, wherein the graph-based connection component technology is applied only to a subset of the sounding data, the subset being based on identified incremental changes from the first set of at least one interference group. [Aspect 4] The CATV system according to aspect 3, wherein the identified incremental changes include pairs of modems within the same IG that no longer interfere with each other. [Aspect 5] The CATV system according to aspect 3, wherein the identified incremental changes include pairs of modems within different IGs that currently interfere with each other but did not interfere with each other at a previous time point. [Aspect 6] The CATV system according to aspect 1, implementing an FDX CATV architecture. [Aspect 7] The CATV system according to aspect 1, wherein the graph-based connection component technology is based on a graph having nodes and edges. [Aspect 8] The CATV system according to aspect 7, wherein the nodes correspond to individual cable modems, and the edges connect the nodes, each having an associated interference measurement criterion measured by a sounding procedure on the CATV system. [Aspect 9] The CATV system according to aspect 8, wherein the graph is undirected. [Aspect 10] The CATV system according to aspect 8, wherein the graph is directed. [Aspect 11] A method of arranging a plurality of modems in a CATV system, the method comprising: automatically arranging the head ends of the plurality of modems in a first arrangement of at least one interference group (IG) by a processor operably connected to a head end of the CATV system; automatically reconfiguring the plurality of modems by a processor operably connected to the head end to a second arrangement of at least one IG different from the first arrangement of the at least one IG, the method comprising: [Aspect 12] The method according to aspect 11, wherein graph-based connection component technology is used to process the array based on the sounding data to reconfigure the plurality of modems to the second arrangement. [Aspect 13] The method according to aspect 12, wherein the graph-based connection component technology is applied only to a subset of the sounding data, the subset being based on identified incremental changes from the first arrangement. [Aspect 14] The method according to aspect 13, wherein the identified incremental changes include pairs of modems within the same IG that no longer interfere with each other. [Aspect 15] The method according to aspect 13, wherein the identified incremental changes include pairs of modems in different IGs that currently interfere with each other and did not interfere with each other at a previous time. [Aspect 16] The method according to aspect 11, wherein an FDX CATV architecture is implemented. [Aspect 17] The method according to aspect 11, wherein the graph-based connection component technology is based on a graph having nodes and edges. [Aspect 18] The method according to aspect 17, wherein the nodes correspond to individual cable modems, and the edges connect the nodes, each having an associated interference measurement criterion measured by a sounding procedure on the CATV system. [Aspect 19] The method according to aspect 18, wherein the graph is undirected. [Aspect 20] The method according to aspect 18, wherein the graph is directed. [Aspect 21] A processing system for exchanging data between a head end and a plurality of cable modems via a transmission network, the processing system comprising: A processor that selectively initiates a periodic baseline sounding test of the cable modem and a periodic full sounding test of the cable modem, wherein the full sounding test generates more data than the baseline sounding test. A memory that stores at least one past baseline sounding test. A processing system in which the processor performs a current baseline sounding test and uses a comparison between the current baseline sounding test and at least one record of the past baseline tests to select whether to initiate a full sounding test. [Aspect 22] The processing system according to aspect 21, wherein the transmission system is full-duplex. [Aspect 23] The processing system according to aspect 21, wherein the full sounding test is one selected from a CW test and an OUDP test. [Aspect 24] The processing system according to aspect 21, wherein the baseline sounding test measures noise at each cable modem when the cable modem is not transmitting data. [Aspect 25] The processing system according to aspect 24, wherein the baseline sounding test generates a record that includes a vector of noise measurements for each band in a plurality of frequency bands across the spectrum of interest. [Aspect 26] Including a database that stores a plurality of past baseline sounding records, and the processor uses one selected record from the plurality of past baseline sounding records to compare with the current baseline sounding test and uses the comparison to select whether to initiate a full sounding test. The processing system according to aspect 21. [Aspect 27] The processing system according to aspect 21, wherein at least one record of the past baseline tests includes at least one of the days of the week for each at least one record and the time of each at least one record. [Aspect 28] The processing system according to aspect 27, wherein at least one record of the past baseline tests includes at least one weather measurement criterion associated with each at least one record. [Aspect 29] The processing system according to aspect 28, wherein the at least one measurement criterion includes ambient temperature data that occurred during the baseline sounding test associated with the record. [Aspect 30] The processing system according to aspect 28, wherein the at least one meteorological measurement criterion is obtained from a meteorological database after at least one record of the past baseline test has been created. [Aspect 31] A method for exchanging data between a head end and a plurality of cable modems via a transmission network, the method comprising: Selectively initiating a periodic baseline sounding test of the cable modem and a periodic full sounding test of the cable modem, wherein the full sounding test generates more data than the baseline sounding test; Storing at least one record of a past baseline sounding test in a memory; Performing a current baseline sounding test; Using a comparison between the current baseline sounding test and at least one record of the past baseline test to select whether to initiate a full sounding test. [Aspect 32] The method according to aspect 31, wherein the transmission system is full-duplex. [Aspect 33] The method according to aspect 31, wherein the full sounding test is one of a CW test and an OUDP test. [Aspect 34] The method according to aspect 31, wherein the baseline sounding test measures noise at each cable modem when the cable modem is not transmitting data. [Aspect 35] The method according to aspect 34, wherein the baseline sounding test generates a record that includes a vector of noise measurements for each band in a plurality of frequency bands across a spectrum of interest. [Aspect 36] The method according to aspect 31, comprising storing a plurality of past baseline sounding records, using one selected record from the plurality of past baseline sounding records for comparison with a current baseline sounding test, and using the comparison to select whether to initiate a full sounding test. [Aspect 37] The method according to aspect 31, wherein at least one record of the past baseline test includes at least one of the days of the week of each at least one record and the time of each at least one record. [Aspect 38] The method according to aspect 37, wherein at least one record of the past baseline test includes at least one meteorological measurement criterion associated with each at least one record. [Aspect 39] The method according to aspect 38, wherein the at least one measurement criterion includes ambient temperature data that occurs during the baseline sounding test associated with the record. [Aspect 40] The method according to aspect 38, comprising the step of obtaining the at least one meteorological measurement criterion from a meteorological database after at least one record of the past baseline test has been created.

Claims

1. A CATV system including a headend connected to a plurality of modems via a transmission network, wherein the plurality of modems are arranged within a first set of at least one interference group (IG), and the CATV system is operably connectable to the headend, and includes a processor capable of reconfiguring the plurality of modems using graph-based connection component technology within at least one second set of IGs different from the first set of at least one IG, the processor reconfiguring the plurality of modems based on the first set of at least one IG, The CATV system, wherein the graph-based connection component technology is a technology for processing an array for identifying an IG corresponding to the graph based on a graph having nodes corresponding to the plurality of modems and edges connectable to the modems, and based on sounding data from the plurality of modems, and reconfiguring the plurality of modems into the second set of at least one IG.

2. The CATV system according to claim 1, wherein the graph-based connection component technology is applied only to a subset of the sounding data, and the subset is based on identified incremental changes from the first set of at least one interference group.

3. The CATV system according to claim 2, wherein the identified incremental changes include pairs of modems within the same IG that no longer interfere with each other.

4. The CATV system according to claim 2, wherein the identified incremental changes include pairs of modems in different IGs that currently interfere with each other but did not interfere with each other at a previous time.

5. The CATV system according to claim 1, implementing an FDX CATV architecture.

6. The CATV system according to claim 1, wherein the nodes correspond to individual cable modems, and the edges connect the nodes and each has an associated interference measurement criterion measured by a sounding procedure on the CATV system.

7. The CATV system according to claim 6, wherein the graph is undirected.

8. The CATV system according to claim 6, wherein the graph is directed.

9. A method of arranging a plurality of modems in a CATV system, the method comprising: A processor operably connected to the headend of the CATV system arranges the plurality of modems connected to the headend in a first arrangement of at least one interference group (IG). The processor operably connected to the headend uses graph-based connection component technology to reconfigure the plurality of modems from the first arrangement to a second arrangement of at least one IG that is different from the first arrangement of the at least one IG. The graph-based connection component technology processes an array that identifies an IG corresponding to the graph based on the plurality of modems and the sounding data from the plurality of modems, which has nodes corresponding to the plurality of modems and edges connectable to the modems, to reconfigure the plurality of modems into the second set of the at least one IG.

10. The method according to claim 9, wherein the graph-based connection component technology is applied only to a subset of the sounding data, and the subset is based on the identified incremental changes from the first arrangement.

11. The method according to claim 10, wherein the identified incremental changes include pairs of modems within the same IG that no longer interfere with each other.

12. The method according to claim 10, wherein the identified incremental changes include pairs of modems in different IGs that currently interfere with each other but did not interfere with each other at a previous time.

13. The method according to claim 9, which implements an FDX CATV architecture.

14. The method according to claim 9, wherein the nodes correspond to individual cable modems, and the edges connect the nodes and each has an associated interference measurement criterion measured by a sounding procedure on the CATV system.

15. The method according to claim 14, wherein the graph is undirected.

16. The method according to claim 14, wherein the graph is directed.

Images