Adaptive Random Access Protocol and Dynamic Search Tree Expansion Resolution for Multiple Station Networks

The present invention relates to a protocol and a dynamic search tree expansion procedure for a communication system in which multiple stations share access to a common communication channel.

Communication systems typically include a plurality of stations connected to one another over a common communication channel. For example, in a community antenna television (CATV) system, a headend is connected by a cable to a plurality of subscriber stations. The cable supports downstream communication from the headend to the subscriber stations and upstream communication from the subscriber stations to the headend. Data, which is transmitted between the headend and the subscriber stations, is transmitted in data frames. Accordingly, when the headend communicates with a subscriber station, the headend transmits in a downstream data frame to the subscriber station, and when a subscriber station communicates with the headend, the subscriber station transmits in an upstream data frame to the headend.

In such a CATV system, the headend and the subscriber stations must share the resources of the cable in some manner. For example, downstream and upstream communications are typically allocated to different frequency ranges. In a sub-split allocation, downstream communications are allocated to a frequency range between 54 MHz and 750 MHz and above, while the upstream communications are allocated to a frequency range below 42 MHz. In a mid-split allocation, downstream communications are allocated to a frequency range of 162 MHz and above, while upstream communications are allocated to a frequency range between 5 to 100 MHz. In a high-split allocation, downstream communications are allocated to a frequency range of 234 MHz and above, while upstream communications are allocated to a frequency range between 5 MHz and 174 MHz.

Moreover, the subscriber stations must also share the resources of the cable in some manner. In a time division multiple access (TDMA) CATV system, the subscriber stations typically share the cable by transmitting data during uniquely assigned and non-overlapping time periods. In a frequency division multiple access (FDMA) CATV system, the subscriber stations share the cable by dividing up the available upstream frequency bandwidth into numerous narrow frequency channels and by allocating to each subscriber station its own corresponding narrow frequency band. In a code division multiple access (CDMA,) CATV system, the subscriber stations share the cable by multiplying their data messages by corresponding assigned code words and then transmitting the result.

A TDMA system, which assigns each subscriber station to a unique time slot, avoids collisions of data transmitted by the subscriber stations but restricts the amount of data throughput from the subscriber stations to the headend. An FDMA system, which allocates to each subscriber station its own corresponding narrow frequency band, similarly restricts throughput because the number of frequency bands allocated to the subscriber stations is limited. A CDMA system likewise limits throughput over a communication cable because the number of code words which are available to be assigned to subscriber stations is limited.

In order to increase throughput of the data transmitted by the subscriber stations to the headend of a CATV system, it is known to divide the upstream data frame, which supports communication from the subscriber stations to the headend, into a number of minislots and data slots. Those subscriber stations, which have data to transmit to the headend, are required to insert a reservation request in a minislot of the current upstream data frame (i.e., the upstream data frame at discrete time n). This reservation request requests the headend to reserve data slots in a subsequent upstream data frame (e.g., the upstream data frame at discrete time n+1) for use by those subscriber stations.

Because the number of minislots in an upstream data frame utilized by such current systems is limited, contention between subscriber stations for access to the is limited number of minislots results in frequent collisions (i.e., contention) between reservation requests. However, it is hoped that at least some reservation requests will be successfully transmitted by subscriber stations to the headend without collision in any given upstream data frame. Therefore, it is generally thought that eventually all subscriber stations will be able to transmit their data to the headend in data slots reserved for that purpose. Yet, because the number of minislots in such systems is limited and fixed, the throughput in such a system is correspondingly limited.

It is also known to increase throughput of the data transmitted by the subscriber stations to the headend of a CATV system by allowing the subscriber stations to contend for the same frequency, time, or code word slots in the upstream data frame and to use a tree algorithm to resolve any resulting collisions. Collisions result when two or more subscriber stations transmit data in the same slot or slots of an upstream data frame. When the headend detects such collisions, the headend initiates a tree algorithm which assigns, for a second layer of contention, a predetermined number of expansion slots in the next upstream data frame for each contention slot in the previous upstream data frame in which there was a collision. In other words, the tree algorithm expands each contention slot in which there was a collision to a predetermined number of expansion slots.

In response to this expansion by the headend, each subscriber station determines whether it transmitted in one of the slots in which there was a collision. Each subscriber station, which determines that it transmitted in a contention slot in which there was a collision, randomly selects in the next upstream data frame one of the expansion slots corresponding to the contention slot of the previous upstream data frame in which it transmitted data. These subscriber stations then re-transmit their data in the corresponding randomly selected expansion slots. If the headend again detects collisions, the tree algorithm assigns, for a third layer of contention, a predetermined number of expansion slots in the next upstream data frame for each slot in the previous upstream data frame in which there was a collision. The contending subscriber stations respond as before. In this way, contention is resolved.

The present invention relates to system which combines the attributes of a protocol which dynamically varies the number of minislots in an upstream data frame, as needed, with the attributes of a dynamic tree algorithm for resolving contention when two or more stations collide in the same slot of an upstream data frame.

In accordance with one aspect of the present invention, a station comprises a receiving means, a transmission parameter generating means, and a transmitting means. The receiving means receives a downstream data frame having a range parameter R and a parameter MAP. The parameter MAP defines a number of new message minislots NMS, a number of expansion minislots EMS expanded by a dynamic search tree expansion procedure, and a number of data slots DS in a next upstream data frame. The transmission parameter generating means generates a transmission parameter N. The transmission parameter N is constrained by the range parameter R. The transmitting means (i) transmits a reservation request in a new message minislot of the next upstream data frame if the transmission parameter N corresponds to the new message minislot, (ii) re-transmits a reservation request in an expansion minislot of the next upstream data frame if the station had transmitted a reservation request in a minislot which was in a previous upstream data frame and which was expanded in the next upstream data frame, and (iii) transmits data in a data slot, if any, reserved to the station.

In accordance with another aspect of the present invention, a station for transmitting data in slots of upstream data messages over a communication medium comprises a receiving means, first and second slot selecting means, and an inserting means. The receiving means receives a downstream data message. The downstream data message includes a slot parameter. The slot parameter indicates new message slots and expansion slots within which reservation requests may be transmitted. The expansion slots result from a dynamic search tree expansion procedure. The first slot selecting means selects one of the expansion slots. The second slot selecting means selects, on at least a pseudorandom basis, one of the new message slots. The inserting means inserts a previously transmitted reservation request in the selected one of the expansion slots of a designated future upstream message and inserts a new reservation request in the selected one of the new message slots of a future upstream message.

In accordance with yet another aspect of the present invention, a method of transmitting data in slots of upstream data messages comprises the following steps: a) receiving a downstream data message, wherein the downstream data message includes a range parameter R and a slot parameter, wherein the slot parameter indicates new message slots NS and expansion slots ES in which reservation requests may be transmitted, and wherein the expansion slots ES result from a dynamic search tree expansion procedure; b) selecting one of the expansion slots; c) selecting one of the new message slots according to the range parameter R; d) inserting a previously transmitted reservation request in the selected one of the expansion slots; and, e) inserting a new reservation request in the selected one of the new message slots.

In accordance with a further aspect of the present invention, a station for transmitting and receiving data over a communication medium comprises a slot expanding means, a slot parameter determining means, a range parameter determining means, and an inserting means. The slot expanding means expands minislots, which are in a data message received by the station and which experienced collision, to expansion slots according to a dynamic search tree expansion procedure. The slot parameter determining means determines a slot parameter for a future upstream data message. The slot parameter indicates (i) new message slots within which new reservation requests may be transmitted to the station, and (ii) the expansion slots. The range parameter determining means determines a range parameter R for a future upstream data message. The range parameter R is determined on the basis of loading of a communication medium. The inserting means inserts the slot parameter and the range parameter R in a downstream data message for communication over the communication medium.

These and other features and advantages of the present invention will become more apparent from a detailed consideration of the invention when taken in conjunction with the drawings in which:

FIG. 1 is a schematic diagram of a CATV system which includes a headend connected to a plurality of subscriber stations by way of a cable and which is exemplary of a communication system configured in accordance with the present invention;

FIG. 2 illustrates a downstream data frame in which the headend transmits to the subscriber stations of FIG. 1;

FIG. 3 illustrates an upstream data frame in which the subscriber stations transmit to the headend of FIG. 1;

FIG. 4 illustrates an example of minislots in upstream data frames F1-F7 according to a dynamic expansion search tree expansion algorithm;

FIG. 5 illustrates a slot format of the upstream data frame;

FIGS. 6A and 6B illustrate a program which is executed by the headend of FIG. 1 in an exemplary implementation of the present invention;

FIGS. 7-10 are graphs useful in explaining the program of FIGS. 6A and 6B; and,

FIGS. 11A and 11B illustrate a program which is executed by each of the subscriber stations of FIG. 1 in the exemplary implementation of the present invention.

FIG. 1 illustrates a CATV system 10 which includes a headend 12, a plurality of subscriber stations 14.sub.1 -14.sub.n, and a cable 16 which interconnects the headend 12 and the subscriber stations 14.sub.1 -14.sub.n. The headend 12 may be of conventional hardware design incorporating a processor which may be programmed to support downstream communication over the cable 16 in accordance with the present invention. Similarly, the subscriber stations 14.sub.1 -14.sub.n may be of conventional hardware design each incorporating a processor which may be programmed to support upstream communication over the cable 16 in accordance with the present invention.

According to the present invention, when the subscriber stations 14.sub.1 -14.sub.n have data to communicate to the headend 12 over the cable 16, those subscriber stations first make a reservation request. In making a reservation request, the subscriber stations 14.sub.1 -14.sub.n contend with one another for a limited but dynamically variable number of minislots in the upstream data frame because minislots carry the reservation requests from the subscriber stations 14.sub.1 -14.sub.n to the headend 12. With respect to those reservation requests which are successfully received by the headend 12 (i.e., received by the headend 12 in minislots in which there are no collisions), the headend 12 acknowledges the reservation request by way of an acknowledgement transmitted by the headend 12 to the successful subscriber stations 14.sub.1 -14.sub.n. Accordingly, the upstream data frame is divided into slots S some or most of which are subdivided into minislots MS such that all contention and reservation activity takes place in the minislots of the upstream data frame and all data transmission takes place in un-subdivided data slots DS of the slots S.

The arrangement of minislots MS and data slots DS is specified by the headend 12 in the downstream data frame. The subscriber stations 14.sub.1 -14.sub.n use this specified arrangement and certain other parameters in order to make a decision as to whether they can transmit reservation requests and data in a subsequent upstream data frame. Accordingly, the subscriber station 14.sub.1 -14.sub.n contend for the bandwidth of the upstream channel. A successful reservation request results in the headend 12 reserving one or more data slots, based on availability, to each of the subscriber stations which had transmitted successful reservation requests. The number of reserved data slots which are available to any one subscriber station depends on the number of subscriber stations making successful reservation requests.

The sizes of the upstream and downstream data frames may be equal, may be fixed, and may be defined, for example, to have a size equal to at least the sum of the headend processing time and the round trip transmission delay of the cable 16.

An exemplary downstream data frame is illustrated in FIG. 2. Each such downstream data frame has four sections. The first section contains a range parameter R. The range parameter R may be used by the subscriber stations 14.sub.1 -14.sub.n to contend for new message minislots when the contending subscriber stations have non-previously transmitted reservation requests to transmit.

The second section of the downstream data frame contains a slot parameter MAP. The slot parameter MAP defines which slots of the next upstream data frame are (i) new message minislots (NMS) which are used by the subscriber stations 14.sub.1 -14.sub.n to transmit new reservation requests (i.e., reservation requests that have not been previously transmitted), (ii) expansion minislots (EMS) which are used by the subscriber stations 14.sub.1 -14.sub.n to transmit reservation requests which were previously transmitted and which collided with reservation requests from other subscriber stations, and (iii) data slots (DS) which are reserved to the subscriber stations 14.sub.1 -14.sub.n so that the subscriber stations 14.sub.1 -14.sub.n can transmit data. The slot parameter MAP, for example, may be a map which defines the positions of each new message minislot, each expansion minislot, and each data slot in the next upstream data frame. The map allows these new message minislots, expansion minislots, and data slots to be interspersed throughout the upstream data frame. The subscriber stations 14.sub.1 -14.sub.n read the map in order to determine which slots in the upstream data frames are new message minislots, which slots in the upstream data frames are expansion minislots, and which slots in the upstream data frames are data slots.

Alternatively, the slot parameter MAP may simply be boundaries which divide the slots in the upstream data frames between new message minislots, expansion minislots, and data slots. In this case, the new message minislots, the expansion minislots, and the data slots are segregated from one another.

In addition, minislots, which are contained in the next upstream data frame, may be assigned an RQ# (i.e., an RQ number). Thus, the new message minislots, which are contained in the next upstream data frame and which are to be used by the subscriber stations 14.sub.1 -14.sub.n to transmit new reservation requests (i.e., reservation requests that have not been previously transmitted), may be assigned an RQ#=0. Thus, a subscriber station, which has a previously untransmitted reservation request to transmit, may transmit the previously untransmitted reservation request in a minislot which is assigned an RQ number of 0.

An RQ number other than 0 may be assigned (i) to expansion minislots that are used by the subscriber stations 14.sub.1 -14.sub.n to re-transmit reservation requests which collided with reservation requests from other subscriber stations in a previous upstream data frame, and (ii) to the contention minislot of the previous upstream data frame to which the expansion minislots correspond. For example, if collisions occurred in minislots 16, 27, 33, and 45 of the previous upstream data frame n, the headend 12 indicates in the slot parameter MAP that previous minislot 16 corresponds to an RQ number of 4, and the headend 12 assigns an RQ number of 4 to the expansion minislots which are in the next upstream data frame n+1 and to which previous minislot 16 is expanded. Similarly, (i) the headend 12 indicates in the slot parameter MAP that previous minislot 27 corresponds to an RQ number of 3, and the headend 12 assigns an RQ number of 3 to the expansion minislots which are in the next upstream data frame n+1 and to which previous minislot 27 is expanded, (ii) the headend 12 indicates in the slot parameter MAP that previous minislot 33 corresponds to an RQ number of 2, and the headend 12 assigns an RQ number of 2 to the expansion minislots which are in the next upstream data frame n+1 and to which previous minislot 33 is expanded, and (iii) the headend 12 indicates in the slot parameter MAP that previous minislot 45 corresponds to an RQ number of 1, and the head end 12 assigns an RQ number of 1 to the expansion minislots which are in the next upstream data frame n+1 and to which previous minislot 45 is expanded.

Accordingly, if a subscriber station has transmitted a reservation request in minislot 16 of the previous upstream data frame n, that subscriber station detects that the minislot 16 of the previous upstream data frame n is assigned an RQ number of 4, and will re-transmit the reservation request in one of the expansion minislots which are assigned an RQ number of 4 and which are in the next upstream data frame n+1. It should be noted that the largest RQ number is determined by the number of minislots experiencing collisions. In the example above, four minislots (i.e., minislots 16, 27, 33, and 45) experienced collisions and, therefore, the largest RQ number is 4.

The third section of the downstream data frame is devoted to the acknowledgement of the reservation requests that are successfully made by the subscriber stations 14.sub.1 -14.sub.n to the headend 12 in a previous upstream data frame. Each acknowledgement field (ACK) may include, for example, (i) a station ID, which is the address of the subscriber station to which an acknowledgement is given, and (ii) the data slot or data slots which are reserved to the subscriber station that is identified by the station ID and in which the identified subscriber station may transmit data to the headend 12.

The fourth section of the downstream data frame contains data slots which may be used by the headend 12 to communicate other data to the subscriber stations 14.sub.1 -14.sub.n.

An example of an upstream data frame is illustrated in FIG. 3. In the upstream channel, the subscriber stations 14.sub.1 -14.sub.n use the slot parameter MAP of the previous downstream data frame in order to determine the definition of the next upstream data frame. The upstream data frame contains a plurality of slots. As shown in FIG. 3, a slot may be subdivided into new message minislots NMS, a slot may be subdivided into expansion minislots EMS, or a slot may be un-subdivided (or partially divided) and used as a data slot DS. The mix of new message minislots NMS, expansion minislots EMS, and data slots DS is defined by the headend 12 and varies depending upon load conditions. Accordingly, the number of minislots changes dynamically dependent upon the extent of collisions and of reservation requests in a reservation request queue DQ at the headend 12. A slot may be subdivided into a fixed number m of new message minislots NMS and/or expansion minislots EMS.

When two or more reservation requests from corresponding subscriber stations collide in a new message minislot or in an expansion minislot of an upstream data frame, that new message minislot or expansion minislot is expanded into a number of expansion minislots in a subsequent upstream data frame. FIG. 4 illustrates a dynamic search tree expansion procedure that is used to expand the new message minislots and/or the expansion minislots in which collisions occurred. With no collisions in a previous upstream data frame, all minislots in the next upstream data frame are new message minislots, and the subscriber stations 14.sub.1 -14.sub.n are allowed to contend for any of these new message minislots. When collisions then occur, the system enters a first layer of contention. Thus, the first layer of contention occurs when there are collisions in new message minislots. The upstream data frame F1 in FIG. 4 shows a first layer of contention. The new message minislots in which no collisions occurred are labelled NC, and the new message minislots in which collisions occurred are labelled C followed by a lower case letter to distinguish between such new message minislots.

Accordingly, the headend 12 expands each of the new message minislots in which collisions occur during the first layer of contention by an expansion coefficient E (which may be six, for example). This value for the expansion coefficient E is dependent upon the amount of collision in the new message minislots of the upstream data frame F1. Thus, the headend 12 expands each of the new message minislots in which collisions occurred to six corresponding expansion minislots for a second layer of contention. The second layer of contention occurs when there are collisions in expansion minislots. For example, the new message minislot Ca is expanded to six expansion minislots with the result that, in the second layer of contention, all of these six expansion minislots contain no collisions. Similarly, the minislot Cb is expanded to six expansion minislots with the result that, in the second layer of contention, one of these six expansion minislots (Cb1) contains collisions, and the other five of these six expansion minislots contain no collisions.

It is noted that, because of a limited number of minislots in an upstream data frame, there may be occasions in which not all of the minislots of the first layer of contention may be expanded into the same next upstream data frame. Accordingly, for example, the minislot Cf from the upstream data frame F1 is expanded to six expansion minislots for the second layer of contention in an upstream data frame F3 instead of in the upstream data frame F2. One of these six expansion minislots (Cf1) contains collisions, and the other five of these six expansion minislots contain no collisions.

It is also noted that an upstream data frame may contain expansion minislots of different layers of contention. For example, the minislot Cg from the upstream data frame F1 representing a first layer of contention is expanded to six expansion minislots in the upstream data frame F3, whereas the minislot Cb1 from the upstream data frame F2 representing a second layer of contention is expanded to three expansion minislots also in the upstream data frame F3.

FIG. 4 illustrates that the expansion coefficient E as set by the headend 12 may be varied between layers of contention. Thus, if the amount of collisions in the second layer of contention is reduced sufficiently, the value of E can be reduced for the second layer of contention.

FIG. 5 illustrates the data which is inserted into a minislot or data slot by a subscriber station. This data includes a source address, a control field, a payload, and error checking data. The source address is the address of the sending subscriber station. The control field indicates the type of message (for example, reservation request or data) transmitted by the subscriber station. The payload field contains either the number of data slots that the subscriber stations is requesting to be reserved in the case of a reservation request message or the data to be transmitted by the subscriber station in the case of a data message. The CRC field contains error checking information.

In determining (i) the number of new message minislots NMS for the next upstream data frame, (ii) the number of expansion minislots EMS for the next upstream data frame, (iii) the expansion coefficient E, and (iv) the range parameter R which is used by the subscriber stations 14.sub.1 -14.sub.n to determine if they can transmit reservation requests in new message minislots, the headend 12 executes a program 100 which is illustrated in FIGS. 6A and 6B.

The program 100 is entered each time that an upstream data frame is received by the headend 12. When the program 100 is entered, a block 102 receives an upstream data frame, and a block 104 stores any reservation requests in this data frame in a reservation request queue DQ at the current time n. The block 104 also stores the number of empty new message minislots, the number of empty expansion minislots, the number of collision new message minislots (i.e., new message minislots in which reservation requests collided), the number of collision expansion minislots (i.e., expansion minislots in which reservation requests collided), the number of successful new message minislots (i.e., new message minislots containing single reservation requests), and/or the number of successful expansion minislots (i.e., expansion minislots containing single reservation requests).

Thereafter, a block 106 determines whether the CATV system 10 is in steady state. When the CATV system 10 is in steady state, the number of reservation requests DQ(n) stored in the reservation request queue DQ at the current discrete time n is greater than the number of data slots DS(n) in the data frame just received by the block 102, but is less than this number of data slots DS(n) multiplied by a constant .alpha.. The constant .alpha. may be 1.6, for example. The upstream data frame just received is designated herein as upstream data frame n and is received at discrete time n. If the system is in steady state, a block 108 determines the number of minislots MS to be allocated to the next upstream data frame n+1 according to the following equation: ##EQU1## wherein S is the total number of slots in a data frame, m is the number of minislots into which a slot may be subdivided, e is 2.718281828 . . . , MS(n+1) is the number of minislots for the next upstream data frame, k is the average number of data slots reserved by reservation requests, and M is the steady state number of minislots. Because a whole slot is used as a data slot, the number of data slots in the next upstream data frame n+1 is, accordingly, given by the following equation: ##EQU2##

If the block 106 determines that the CATV system 10 is not in steady state, a block 110 determines whether the number of reservation requests DQ(n) in the reservation request queue DQ of the headend 12 at discrete time n is less than the number of data slots DS(n) in the upstream data frame just received. If so, a block 112 determines the number of minislots MS to be allocated to the next upstream data frame n+1 according to the following equation:

wherein DQ(n) is the number of reservation requests DQ(n) in the reservation request queue DQ of the headend 12 at time n. The number of data slots DS(n+1) in the next upstream data frame n+1, accordingly, is set to DQ(n).

If the block 106 determines that the CATV system 10 is not in steady state, and if the block 110 determines that the number of reservation requests DQ(n) in the reservation request queue DQ of the headend 12 at discrete time n is not less than the number of data slots DS(n) in the upstream data frame just received, a block 114 determines the number of minislots MS to be allocated to the next upstream data frame n+1 according to the following equation: ##EQU3## where DS(n) is the number of reserved data slots in the upstream data frame just received. The number of data slots DS(n+1) in the next upstream data frame n+1 is, accordingly, given by the following equation: ##EQU4## As is described below, the minislots MS(n+1) will be divided between new message minislots and/or expansion minislots.

A block 116 analyzes the upstream data frame received by the block 102 in order to determine which non-empty new message minislots therein contain non-contending reservation requests. These new message minislots contain reservation requests from only one subscriber station. The block 116 sets a parameter SUCN equal to the number of these minislots. The block 116 also analyzes the upstream data frame received by the block 102 in order to determine which non-empty expansion minislots therein contain non-contending reservation requests. These expansion minislots contain reservation requests from only one subscriber station. The block 116 sets a parameter SUC.sub.E equal to the number of these minislots. The block 116 may, if desired, determine a parameter SUC, for each layer of contention subsequent to the first layer of contention so that a parameter SUC.sub.E1 pertains to the second layer of contention, a parameter SUC.sub.E2 pertains to the third layer of contention, and so on.

A block 118 analyzes the received upstream data frame to determine which non-empty new message minislots therein contain contending reservation requests. These new message minislots contain reservation requests from more than one subscriber station. The block 118 sets a parameter COL.sub.N equal to the number of these new message minislots. The block 118 also analyzes the received upstream data frame to determine which non-empty expansion minislots therein contain contending reservation requests. These expansion minislots contain reservation requests from more than one subscriber station. The block 118 sets a parameter COL.sub.E equal to the number of these expansion minislots. The block 118 may, if desired, determine a parameter COL.sub.E for each layer of contention subsequent to the first layer of contention so that a parameter COL.sub.E1 pertains to the second layer of contention, a parameter COL.sub.E2 pertains to the third layer of contention, and so on.

A block 120 chooses whether to calculate the number of active stations, N, based upon the parameter SUC. If so, a block 122 determines the number of active stations N from the following equation: ##EQU5## wherein N is the number of active stations, wherein MS is a total number of minislots in the upstream data frame just received, and wherein SUC is the number of non-empty minislots in which collisions did not occur as determined by the block 116. The block 120 may determine a number of active stations N for each layer of contention so that a number of active stations N.sub.N is determined from the parameter SUC.sub.N, a number of active stations N.sub.E1 is determined from the parameter SUC.sub.E1, a number of active stations N.sub.E2 is determine from the parameter SUC.sub.E2, and so on.

Alternatively, the number of active stations N may be determined by the block 122 from a lookup table which is stored in memory at the headend 12 according to the graph shown in FIG. 7. This graph corresponds to equation (4). The vertical axis of this graph is the input axis. The parameter SUC as determined by the block 104 is input along the vertical axis. The horizontal axis is the output axis along which the number of active stations N is determined as a function of the input vertical axis. When using the graph of FIG. 7, an ambiguity arises because there are two output values along the horizontal axis for each input along the vertical axis. This ambiguity may be resolved by the use of the collision curve shown in FIG. 8. This curve of FIG. 8 is explained in more detail below.

On the other hand, if the block 120 chooses not to calculate the number of active stations based upon the parameter SUC, a block 124 determines the number of active stations N from the following equation: ##EQU6## wherein N is the number of active stations, wherein MS is a total number of minislots in the upstream data frame just received, and wherein COL is the number of minislots, which are in an upstream data frame just received and in which collisions occurred, as determined by the block 118. The block 124 may determine a number of active stations N for each layer of contention so that a number of active stations N.sub.N is determined from the parameter COL.sub.N, a number of active stations N.sub.E1 is determined from the parameter COL.sub.E1, a number of active stations N.sub.E2 is determine from the parameter COL.sub.E2, and so on.

Alternatively, the number of active stations N may be determined from a lookup table which is stored in memory at the headend 12 according to the graph shown in FIG. 8. This graph corresponds to equation (5). The vertical axis of this graph is the input axis. The parameter COL as determined by the block 118 is input along the vertical axis. The horizontal axis is the output axis along which the number of active stations N is determined as a function of the vertical axis.

The block 120 may make its decision based upon a user set flag or upon other criteria.

It should be apparent that the CATV system 10 can be arranged to determine the number of active stations N exclusively from equation (4). If so, the block 122 must use the collision parameter COL and the block 124 in order to resolve the ambiguity described above. Alternatively, the CATV system 10 can be arranged to determine the number of active stations N exclusively from equation (5), in which case there is no ambiguity to resolve. If so, the blocks 116, 120, and 122 may be eliminated. As a further alternative, the number of active stations N can be determined from a combination of the parameter SUC and the parameter COL.

A block 126 uses the number of active stations N in order to determine the expansion parameter E according to a lookup table which is stored in memory at the headend 12 and which conforms to the graph shown in FIG. 9. The horizontal axis of this graph is the input axis. The number of active stations N, as determined by one of the methods described above, is input along the horizontal axis. The vertical axis is the output axis along which the expansion parameter E is determined as a function of the horizontal axis. Accordingly, an expansion coefficient E.sub.N is determined based upon the number of active stations N.sub.N for the first layer of contention, an expansion coefficient E.sub.E1 is determined based upon the number of active stations N.sub.E1 for the second layer of contention, an expansion coefficient E.sub.E2 is determined based upon the number of active stations N.sub.E2 for the third layer of contention, and so on.

Alternatively, the block 126 may determine the expansion parameter E according to a lookup table which is stored in memory at the headend 12 and which conforms to the graph shown in FIG. 10. The horizontal axis of this graph is the input axis. The parameter COL as determined by the block 118 is input along the horizontal axis. The vertical axis is the output axis along which the expansion parameter E is determined as a function of the horizontal axis.

It should be recognized that the blocks 116-126 determine the expansion coefficient E for each corresponding layer of contention as a dynamically variable expansion coefficient such as that described in connection with FIG. 4.

A block 128 determines the number of expansion minislots in the next upstream data frame n+1 as the parameter EMS(n+1), and the number of new message minislots in the next upstream data frame n+1 as the parameter NMS(n+1). The block 128 determines an initial parameter EMS.sub.i (n+1) by expanding the new message minislots and expansion minislots in the upstream data frame n in which collisions occurred according to the corresponding expansion coefficients E. That is, the block 128 expands the new message minislots in the upstream data frame n in which collisions occurred (if any) by the expansion coefficient E.sub.N, the block 128 expands the first layer expansion minislots in the upstream data frame n in which collisions occurred (if any) by the expansion coefficient E.sub.E1, the block 128 expands the second layer expansion minislots in the upstream data frame n in which collisions occurred (if any) by the expansion coefficient E.sub.E2, and so on. All of these expanded minislots are added to produce the initial parameter EMS.sub.i (n+1). The block 128 then subtracts the initial parameter EMS.sub.i (n+1) from the number of minislots MS to be allocated to the next upstream data frame n+1 as determined by the blocks 106-114. If the result is less than NMS.sub.min (which may be set to zero, four, or any other number), the block 128 sets EMS(n+1) equal to MS-NMS.sub.min, and sets NMS(n+1) equal to NMS.sub.min, where MS is determined by the blocks 106-114. If the result is not less than NMS.sub.min, the block 128 sets EMS(n+1) equal to EMS.sub.i (n+1), and sets NMS(n+1) equal to MS-EMS.sub.i (n+1). In this way, an upstream data frame contains no less than NMS.sub.min new message minislots.

A block 130 determines the range parameter R from the parameter COL according the following equation: ##EQU7## where n indicates the current frame, n+1 indicates the next upstream data frame, R(n+1) is the range parameter for the next upstream data frame n+1, R(n) is the range parameter for the upstream data frame n just received, N represents the number of active stations N.sub.N as determined by some combination of the blocks 116-124, NMS(n+1) is the number of new message minislots in the next upstream data frame n+1 as determined by the block 128, NMS(n) is the number of new message minislots in the upstream data frame n just received, COL(n) is the parameter COL as determined by the block 118 based upon the upstream data frame n just received, and e is 2.718281828 . . . .

A block 132 determines the slot parameter MAP. If the slot parameter MAP is a map of the new message minislots, expansion minislots, and data slots, the block 132 constructs the map based upon NMS(n+1) and EMS(n+1) as determined by the block 128 and upon DS as described above in connection with the blocks 106-114, according to any desired rule. Alternatively, the block 132 determines the slot parameter MAP by allotting a first portion of the upstream data frame n+1 to the NMS(n+1) new message minislots, by allotting a next portion of the upstream data frame n+1 to the EMS(n+1) expansion minislots, and by allotting a remaining portion of the upstream data frame n+1 to the DS data slots. The block 132 also assigns RQ numbers and minislot designations as discussed above. The block 132 inserts the slot parameter MAP, and inserts the range parameter R(n+1) as the parameter R, into the next downstream data frame to be transmitted.

A block 134 inserts any additional information into the next downstream data frame and transmits the next downstream data frame over the cable 16. Thereafter, the program 100 returns to the block 102 to await the next upstream data frame.

Each of the subscriber stations 14.sub.1 -14.sub.n executes a program 200 as shown in FIG. 11A. When the program 200 is entered, a block 202 causes its corresponding subscriber station to wait for the next downstream data frame which contains, inter alia, the range parameter R, the slot parameter MAP, and acknowledgements (which include reserved data slot allocations to the subscriber station). When the next downstream data frame is received, a block 204 determines from the slot parameter MAP whether its corresponding subscriber station, and one or more other subscriber stations, transmitted reservation requests in the same minislot of the previous upstream data frame, i.e., whether the reservation request transmitted by its corresponding subscriber station collided with a reservation request transmitted by one or more other subscriber stations. The block 204 can make this determination, for example, by comparing the minislot in which its corresponding subscriber station transmitted data in the previous upstream data frame to the minislots that are expanded in the slot parameter MAP (e.g., by determining whether the minislot, which was in the previous upstream data frame and in which its subscriber station transmitted a reservation request, has been assigned, in the downstream data frame just received, an RQ number other than 0).

If the block 204 determines that its corresponding subscriber station, and one or more other subscriber stations, transmitted reservation requests in the same minislot of the previous upstream data frame, a block 206 determines, from the slot parameter MAP, whether its corresponding subscriber station is allowed to re-transmit its previously contending data in the next upstream data frame. For example, the block 206 may make this determination by comparing the RQ number, which is assigned to the minislot of the previous upstream data frame in which its subscriber station transmitted its previously contending reservation request, to the RQ numbers assigned to the minislots of the next upstream data frame. If the block 206 does not find a match, the block 206 determines that it is not allowed to re-transmit its previously contending reservation request. On the other hand, if the block 206 finds a match, the block 206 determines that it is allowed to re-transmit its previously contending reservation request.

Accordingly, if the block 206 determines that it is allowed to re-transmit its previously contending reservation request, a block 208 generates a random number N within the range of the expansion minislots assigned to its corresponding subscriber station. That is, the random number N is generated so that it equals one of the expansion minislots corresponding to the minislot in which the corresponding subscriber station transmitted contending data in the previous upstream data frame (i.e., so that the random number N equals one of the expansion minislots having the same RQ number as the RQ number assigned to the minislot in the previous upstream data frame in which the subscriber station, corresponding to the block 206, transmitted its contending reservation request). A block 210 then inserts the previously contending reservation request in the expansion minislot N of the next upstream data frame for transmission. A block 212 resets a COL flag.

It should be noted that, if the block 206 determines that the RQ number, which was assigned to the minislot of the previous upstream data frame in which its subscriber station transmitted its previously contending reservation request, is greater than the largest RQ number assigned to the minislots of the next upstream data frame, the block 208 generates a random number N so that the random number N equals one of the expansion minislots which is in the next upstream data frame and which has the largest RQ number assigned to it. The block 210 then inserts the previously contending reservation request in that expansion minislot N.

On the other hand, the block 206 may determine that its corresponding subscriber station is not permitted to re-transmit its previously contending reservation request in the next upstream data frame (i.e., that its corresponding subscriber station has been assigned a later upstream data frame in which it is allowed to re-transmit its previously contending reservation request). For example, if the RQ number assigned to the minislot, which was in the previous upstream data frame and in which the subscriber station, corresponding to the block 206, transmitted its previously contending reservation request, is smaller than the smallest RQ number assigned to the minislots of the next upstream data frame, then this subscriber station is not permitted to re-transmit its reservation request in this next upstream data frame. In this case, a block 214 sets the COL flag.

If the block 204 determines that its corresponding subscriber station, and one or more other subscriber stations, did not transmit reservation requests in the same minislot of the previous upstream data frame, a block 216 determines whether the COL flag is set. As indicated by the block 214, the COL flag is set when the subscriber station has unsuccessfully transmitted a reservation request but it is not permitted to re-transmit this reservation request in the next upstream data frame. Accordingly, during a subsequent pass through the program 200, the block 216 allows the block 206 to determine if the subscriber station can now re-transmit its previously contending reservation request. If the COL flag is set, program flow passes to the block 206.

If the COL flag is not Bet as determined by the block 216, or after the block 214 sets the COL flag, or after the block 212 resets the COL flag, a block 218 (FIG. 11B) determines whether the corresponding subscriber station (i) has data to transmit and (ii) has not previously transmitted a reservation request for this data. If the corresponding subscriber station (i) has data to transmit and (ii) has not previously transmitted a reservation request for this data, a block 220 generates a transmission parameter RN within the range established by the range parameter R that is contained in the downstream data frame just received from the headend 12. The range established by the range parameter R may be the range between zero and R inclusive, the range between one and R inclusive, or the like. This range contains all of the new message minislots which are assigned an RQ number of 0. The transmission parameter RN is used to determine whether its corresponding subscriber station is permitted to transmit a new reservation request to the headend 12. The transmission parameter RN may be generated by the block 220 as a random number. Thus, because each subscriber station 14.sub.1 -14.sub.n generates its own transmission parameter RN as a random value within the range established by the range parameter R, the transmission probabilities of the subscriber stations 14.sub.1 -14.sub.n are statistically spread along the interval of the range defined by the range parameter R.

A block 222 determines whether the value of N just generated by the block 220 corresponds to one of the new message minislots as defined by the slot parameter MAP contained in the downstream data frame just received from the headend 12. That is, if the slot parameter MAP is a map, the block 222 determines whether the value of N is equal to one of the new message minislots defined in the map. On the other hand, if the slot parameter MAP contains boundaries between the new message minislots, expansion minislots, and data slots, the block 222 determines whether the value of N falls within the new message minislot boundary. If the value of N corresponds to the new message portion of the slot parameter MAP, a block 224 inserts a transmission reservation request in the new message minislot which is equal to the value of N and which is in the upstream data frame being assembled for transmission back to the headend 12.

If the block 222 determines that the value of N does not correspond to the new message portion of the slot parameter MAP, or if the block 218 determines that the subscriber station has no new data for which a reservation request is required, or after the block 224 has inserted a reservation request in the new message minislot having a value equal to N, a block 226 determines whether the downstream data frame just received has a reserved data slot in the next upstream data frame within which the subscriber station may transmit old data. Old data is data for which a previous reservation request had been successfully made by the appropriate subscriber station and the headend 12 has reserved one or more data slots for the data. If so, a block 228 inserts this old data into the data slot reserved by the headend 12 for this subscriber station.

If the block 226 determines that a data slot had not been reserved for the subscriber station in the next upstream data frame, or after the block 228 inserts old data into a data slot reserved in response to a previous reservation request, a block 230 transmits its portion of the next upstream data frame oven the cable 16. Thereafter, the program 200 returns to the block 202 to await the next downstream data frame.

Accordingly, the present invention combines the adaptive allocation of channel resources dependent upon the amount of demand created by the subscriber stations for the upstream channel with a dynamic search tree expansion procedure for resolving contention between transmitting stations. As the number of subscriber stations having data to transmit to the headend 12 increases, the chance that collisions will occur in the upstream data frames also increases. As the number of collisions in the upstream data frames increases, the value of the range parameter R is increased which tends to decrease the probability that a subscriber station will be able insert a reservation request in a new message minislot of subsequent upstream data frames. Moreover, as the number of subscriber stations having data to transmit to the headend 12 increases, the number of reservation requests in the reservation request queue DQ of the headend 12 also increases. As the number of reservation requests in the reservation request queue DQ increases, the number of minislots that are allocated to subsequent upstream data frames decreases.

Similarly, as the number of subscriber stations having data to transmit to the headend 12 decreases, the chance that collisions will occur in the upstream data frames also decreases. As the number of collisions in the upstream data frames decreases, the value of the range parameter R is decreased which tends to increase the probability that a subscriber station will be able insert a reservation request in a new message minislot of subsequent upstream data frames. Moreover, as the number of subscriber stations having data to transmit to the headend 12 decreases, the number of reservation requests in the reservation request queue DQ of the headend 12 also decreases. As the number of reservation requests in the reservation request queue DQ decreases, the number of minislots that are allocated to subsequent upstream data frames increases.

Thus, as the number of reservation requests increases, the headend 12 decreases the number of minislots allocated to subscriber stations in order to reduce the number of successful reservation requests being transmitted by the subscriber stations in the upstream data frames. Also, as the number of collisions in minislots of upstream data frames increases, the value of the range parameter R is increased which has the effect of reducing the number of subscriber stations permitted to transmit reservation requests in the new message minislots which are allocated to subsequent upstream data frames. Accordingly, the slot parameter MAP and the range parameter R work together to adaptively regulate data traffic in the CATV system 10.

At the same time, contention occurring in new message minislots and expansion minislots is resolved using a dynamic tree search procedure. Thus, contention reduction is enhanced and the number of iterations required to reduce contention is lessened, which reduces delay in the transmission of data.

The use of RQ numbers has several advantages. For example, an RQ number of 0 can be assigned to a select group of subscriber stations by use of a selective group address or specific individual addresses. In this way, the probability that certain select subscriber stations can transmit reservation requests may be controlled. Accordingly, a limited number subscriber stations may be assigned an RQ number of 0 to thereby increase the probability that they will successfully transmit reservation requests.

Alternatively, a specific type of subscriber stations may be assigned an RQ number of 0 by use of a selective group address or specific individual addresses in order to thereby allow only that type of subscriber station to successfully transmit reservation requests. Thus, meters could be read during off peak hours when traffic from other types of subscriber stations is usually low.

Certain modifications of the present invention have been discussed above. Other modifications will occur to those practicing in the art of the present invention. For example, as described above, each subscriber station determines its transmission parameter RN as a random number which is only constrained to fall within the range established by the range parameter R. Instead, the transmission parameter RN may be determined by each subscriber station on a pseudorandom basis or on any other basis which tends to spread the transmission parameters RN of the subscriber stations throughout the range R. Therefore, the random generation of the transmission parameter RN should be understood to include not only random generation of the transmission parameter RN but also pseudorandom generation of the transmission parameter RN and generation of the transmission parameter RN by similar methods.

Also, the sizes of the upstream and downstream data frames are described above as being fixed. However, the sizes of the upstream and downstream data frames may be dynamically variable so that the size of these data frames, for example, may be dependent upon traffic load.

In addition, a specific procedure for determining the range parameter R is described above. Instead, the range parameter R may be determined in other ways. For example, the range parameter R may be determined as a function of the successful parameter SUC or as a function of both the collision parameter COL and the successful parameter SUC.

Moreover, as described above, the contention range parameter R is used by the subscriber stations in generating the transmission parameter RN when the subscriber stations have data to transmit, regardless of the priority of that data. Alternatively, a contention range parameter R.sub.L may be used by the subscriber stations in generating one transmission parameter RN.sub.L when the subscriber stations have low priority data to transmit, and a different contention range parameter R.sub.H may be used by the subscriber stations in generating another transmission parameter RN.sub.H when the subscriber stations have high priority data to transmit. The transmission parameter RN.sub.L corresponding to the contention range parameter R.sub.L then determines if the subscriber stations can transmit low priority data, and the transmission parameter RN.sub.H corresponding to the contention range parameter R.sub.H then determines if the subscriber stations can transmit high priority data. Accordingly, the subscriber stations can be given a greater chance of successfully transmitting a reservation request when they have high priority data to transmit.

Also as described above, the cable 16 interconnects the headend 12 and the subscriber stations 14.sub.1 -14.sub.n. However, the headend 12 and the subscriber stations 14.sub.1 -14.sub.n may be interconnected by any communication medium such as a twisted pair, a fiber optic cable, over the air, by way of satellite, and/or the like.

As described above, the number of active stations N can be determined from the parameters SUC and/or COL. However, the number of active stations N can be determined instead from the number of empty of slots (i.e., a parameter EMP) in a previous upstream data frame, or by a combination of the parameters EMP, SUC, and/or COL.

Moreover, as described above, the expansion coefficient E may be re-determined for each layer of contention. Instead, the expansion coefficient E may be determined according to a first level of contention, and may thereafter be reduced by fixed amounts for subsequent layers of contention.

Furthermore, the present invention has been described above in the context of a CATV system. It should be understood, however, that the present invention is useful in a wide variety of communication systems.

Also, as described above, if an upstream data frame is to contain minislots, the minimum number of new message minislots that upstream data frame can have is NMS.sub.min. In order to accommodate this minimum number of new message minislots, it may be desirable to adjust the size of the upstream data frame. It may also be necessary to adjust the size of the upstream data frame if there are not enough expansion minislots in a predetermined number of upstream data frames to efficiently resolve contention.

Moreover, the minimum number of new message minislots NMS.sub.min need not be fixed. Instead, the number of expansion minislots is first determined and is then subtracted from the total number of minislots MS. If the result of this subtraction is that there is one minislot that can be assigned as a new message minislot, then a slot is subdivided by m to produce m+1 new message minislots. Similarly, if the result of this subtraction is that there are two minislots that can be assigned as new message minislots, then a slot is subdivided by m to produce m+2 new message minislots, if the result of this subtraction is that there are three minislots that can be assigned as new message minislots, then a slot is subdivided by m to produce m +3 new message minislots, and so on. A cap may be put on this procedure so that if the result of this subtraction is that there are four minislots that can be assigned as new message minislots, then only four minislots are assigned as new message minislots.

Furthermore, as described above, contention activities take place in new message minislots (NMS) and in expansion minislots (EMS). However, contention activities may alternatively take place in normal slots such as new message slots (NS) and in expansion slots (ES).

Also, as described above, contention activities take place in new message minislots (NMS) and in expansion minislots (EMS). However, contention activities may alter natively take place in normal slots such as new message slots (NS) and in expansion slots (ES). In this case, equation (1) may be rewritten according to the following equation: ##EQU8## wherein CS(n+1) is the total number of contention slots in a data frame, CS=NS+MS, n+1 designates the next upstream data frame, e is 2.718281828 . . . , k is the average number of data slots reserved by reservation requests, and M is the steady state number of contention slots; equation (2) may be rewritten according to the following equation:

wherein DQ(n) is the number of reservation requests DQ(n) in the reservation request queue DQ of the headend 12 at time n; equation (3) may be rewritten according to the following equation: ##EQU9## where DS(n) is the number of reserved data slots in the upstream data frame just received; equation (4) may be rewritten according to the following equation: ##EQU10## wherein N is the number of new active stations, wherein NS is a total number of new message slots in the upstream data frame just received, and wherein SUC is the number of non-empty new message slots in an upstream data frame in which collisions did not occur; and equation (5) may be rewritten according to the following equation: ##EQU11## wherein N is the number of new active stations, wherein NS is a total number of new message slots in the upstream data frame just received, and wherein COL is the number of new message slots which are in an upstream data frame just received and in which collisions occurred. Accordingly, slots as referred to herein includes full slots, partial slots, minislots, or the like unless otherwise specified.

Accordingly, the description of the present invention is to be construed as illustrative only and is for the purpose of teaching those skilled in the art the best mode of carrying out the invention. The details may be varied substantially without departing from the spirit of the invention, and the exclusive use of all modifications which are within the scope of the appended claims is reserved.