JP5594688B2 - Communication system in PTMP type network - Google Patents

Description

  Embodiments described herein relate generally to a communication system in a PTMP type network.

  In recent years, high-speed Internet services using ADSL, FTTH, etc. have rapidly spread mainly in developed countries. In the future, it will become an issue to develop in depopulated areas and developing countries where the population density is low, that is, to develop an access network to eliminate the digital divide. In such a region, the use of a wireless transmission medium such as a microwave band can be considered as one measure for improving the access environment in a short period of time with good economic efficiency. To realize this, it is necessary to develop a new access control method for efficiently sharing a transmission medium in a wide area having a radius of several tens of kilometers.

HUB (hybrid fiber and coaxial) and subscriber wireless access (FWA), such as fixed wireless access (FWA), a single hub device (hereinafter referred to as SCH)
In a point-to-multipoint (PTMP) type access network that is shared by multiple subscriber devices (hereafter referred to as CPE) via a hub, the transmission delay time associated with long-distance transmission increases. Considering this, it is necessary to improve transmission efficiency by sharing resources. From this point of view, for example, DOCSIS (data-over-cable service interface specifications) in HFC (Non-Patent Documents 1 and 2), DAMA (demand assignment multiple access) (Non-Patent Document 3) in satellite communications, etc. DOCSIS is also adopted in IEEE 802.16 (Non-Patent Document 4) that has been proposed and standardized as wireless access (WiMAX).

  In these methods, a reservation type access control method (hereinafter abbreviated as a reservation method) that performs communication after performing a transmission procedure in a reservation cycle of about 2 to 4 msec and establishing a connection is often adopted. There are problems such as that it takes time to establish a connection and it is difficult to respond to a large number of burst transfer requests.

  In contrast, a CSMA-based competitive access control method (hereinafter abbreviated as a competitive method) widely used in Ethernet (registered trademark) (Non-Patent Document 5) and wireless LAN (IEEE802.11x) (Non-Patent Document 6). ), Although there are no drawbacks like the reservation method, most of them limit the maximum network length due to the restriction of the method of “collision detection before frame transmission is completed”, and the throughput characteristics depend on the network length and physical transmission speed. Strongly dependent. For this reason, half-duplex gigabit ethernets are prepared to degrade the throughput characteristics, add dummy data called carrier extension to short frames (eg, add 448 bytes to the shortest frame of 64 bytes) The network length is 100m.

“Data-over-Cable Service Interface Specifications, Cable Modem Termination System Network Side Interface Specification,” SP-CMTS-NSII01-960702, MCNS holdings, L.P., July 1996. “Data-over-Cable Service Interface Specifications, Cable Modem to Customer Premise Equipment Interface Specification,” SP-CMCI101-960702, MCNS Holdings, L.P., July 1996. H. Peyravi, “Medium Access Control Protocols Performance in Satellite Communications,” IEEE Communications Magazine, pp.62-71, March 1999. IEEE Std 802.16-2004, "IEEE Standard for Local and metropolitan area networks Part16: Air Interface for Fixed Broadband Wireless Access Systems," Oct. 2004. IEEE Std 802.3-2002, “Carrier sense multiple access with collision detection (CSMA / CD) access method and physical layer specifications,” 2002. IEEE Std 802.11, “Information technology-Telecommunications and information exchange between systems-Local and metropolitan area networks-Specific requirements-Part11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications,” 1999. Hiroshi Kobayashi, Kazuyoshi Ozawa, Koichiro Kamura, Osamu Moriya, “Synchronous CSMA / CD in PTMP access network,” IEICE (B), Vol.85-B, No.4, pp.471-479, 2002. Osamu Moriya, Hiroshi Kobayashi, "Reduction of collision probability by Synchronous CSMA with Multiple CA in PTMP type access network," A theory of psychology, Vol.44, No.3, pp.932-941, 2003. Osamu Moriya, Hiroshi Kobayashi, “Variation of Collision Avoidance Slot of Synchronous CSMA / MCA Method in PTMP Type Access Network,” Information Science, Vol.44, No.8, pp.2208-2217, 2003. Osamu Moriya, Hiroshi Kobayashi, Satoshi Mori, Takehiro Takahashi, “Application of frame advance transmission control to Synchronous CSMA / v-MCA method in PTMP type access network,” Psychology, Vol.48, No.4, pp. 1758-1766, 2007. Satoshi Mori, Hiroshi Kobayashi, Yoichiro Ueno, Osamu Moriya, “Modeling of Dynamic Fluctuating Traffic Reflecting Service Usage and Development of Simulation Environment,” IEICE Technical Report NS2006-11, pp.67-71, 2006. http://www.creativyst.com/cgi-bin/M/Glos/st/ Get-Term.pl?fsGetTerm=Out of Band (2001).

  Specifically, the conventional wireless LAN (IEEE802.X) has the following problems.

(1) Carry out carrier detection (CS) and CA procedures at the responsibility of the terminal
Since the CSMA / CA method is an access control method that assumes equal distribution, there is a large amount of time that cannot be used for transmission of payloads such as DIFS, SIFS, and back-off-time for each CS and CA. Only a maximum throughput of about 30% to 30% can be obtained.

(2) Hidden terminal problem In order to avoid the hidden terminal problem and the exposed terminal problem, it is necessary to apply a method of confirming the communication with the other party by transmitting RTS / CTS frames before transmitting the payload. Therefore, the throughput must be perceived as degraded.

(3) Non-compliance with PTMP traffic Since access (AP → CPE) from the access point (AP) and upstream access from many CPEs are accessed and compete in an equal position, the Internet and servers The transmission opportunity of a large amount of downstream traffic sent from the network is suppressed, and the effective throughput for the service is deteriorated.

(4) Lack of tolerance during overload
Since only one CA slot is prepared, combined with the unreasonableness of (1) and (2), when a load with a normalized load traffic ρ exceeding 0.3 is applied to the NW, the throughput characteristics deteriorate rapidly and congestion occurs. The system is not stable. For example, it is difficult to get out of this state once it is in a congested state.

(5) Dependence on physical specifications of network characteristics Although there is no problem with the physical specifications of the current wireless LAN (maximum NW length: 100 m, physical transmission speed: 100 Mbps), resource waste time due to RTT is involved for each access control. Therefore, expansion beyond the current specifications is difficult.

According to a first aspect of the present invention, in a hub apparatus in a PTMP network in which a hub apparatus is shared by a plurality of subscriber apparatuses and transmission control is performed between the hub apparatus and the subscriber apparatus, Bands different from the uplink control channel for transmitting CA (collision avoidance) frames with the plurality of subscriber apparatuses and the data channel for transmitting data frames are allocated, and the uplink control channel and the data channel are allocated. Means for changing the transmission capacity of the uplink control channel and the data channel based on the load of the uplink control channel and the data channel when at least one of them is in a high load state. A method for independently performing transmission control and transmission control of data frames of the data channel. A hub device comprising a step.

FIG. 1 is a diagram for explaining the S-CSMA / v-MCA scheme. Figures 2 (a)-(c) set the bandwidth allocation ratio between BW _CA and BW _MAC , assuming a high load condition (ρ = 1 to 2) that consumes almost 100% of the network bandwidth. It is the figure which showed the mode of the channel occupation in SCH about three cases. It is a figure for demonstrating the flow control with respect to CA procedure of this Embodiment. It is a figure for demonstrating operation | movement of the Full-flying S-CSMA / v-MCA system of this Embodiment. FIG. 5 is a diagram showing a configuration of the PTMP access network according to the present embodiment. FIG. 6 is a diagram showing a SCH configuration of the PTMP access network according to the present embodiment. FIG. 7 is a diagram showing a CPE configuration of the PTMP access network according to the present embodiment. FIG. 8 is a diagram illustrating SCH state transition. FIG. 9 is a diagram illustrating state transition of CPE. FIG. 10 is a flowchart for explaining the operation of the SCH. FIG. 11 is a flowchart for explaining the operation of the SCH. FIG. 12 is a flowchart for explaining the operation of the CPE. FIGS. 13A and 13B are diagrams for evaluating the validity of the bandwidth allocation ratio. FIGS. 14A and 14B are diagrams for explaining the effectiveness of flow control. FIG. 15 is a diagram for explaining a change in the number of CA slots. 16A and 16B are diagrams for explaining the characteristics of the conventional NF method and F method and the FF method of the present embodiment.

  The full-flying Synchronous CSMA / v-MCA (carrier sense multiple access with variable multiple collision avoidance analysis) method of this embodiment takes advantage of the characteristics of the PTMP network and avoids collision by aiming at the carrier disappearance time on the upstream transmission medium. (Hereafter, abbreviated as CA) By trying to send all the CA frames for procedures and MAC frames for payload transmission in advance (flying), the use efficiency of the transmission medium will be increased as much as possible. Note that “preceding (flying)” in the present embodiment means close to the run-up oscillation used in bicycle competitions and does not mean so-called flying (premature start) in athletics. As a result, while maintaining the low delay characteristic that is the feature of the competitive system, it is possible to realize a high throughput characteristic similar to the reservation system without being affected by the network length or physical transmission speed of several tens of kilometers.

  The following summarizes the outline of the synchronous CSMA / v-MCA system (Non-Patent Documents 7 to 10) and the problems in developing the system that are the basis of this embodiment, and then details the details of this embodiment to the theoretical performance limits. This will be explained while showing performance evaluation by simulation.

[ S-CSMA / v-MCA method ]
・ PTMP type network and Synchronous CSMA / v-MCA method
A PTMP network that shares a separate transmission medium (upstream (CPE → SCH) and downstream (SCH → CPE)) between a single SCH and multiple CPEs, such as Ethernet based on equal distributed control It has features not found in LAN. In other words, in the equally distributed access method, each CPE performs the carrier detection and collision detection on its own responsibility, but in the PTMP type, the SCH performs carrier detection and collision detection on the upstream transmission medium instead of the CPE, and further performs the MAC. Transmission timing of frames and the like can be controlled with high accuracy.

FIG. 1 is a diagram for explaining an S-CSMA / v-MCA scheme using this feature. In this method,
(1) Bit-synchronize the arrival of the upstream frame on the SCH at a predetermined timing, and make the received signal level on the SCH equal.
(2) The SCH determines the number of CA slots (CA frame reception time slots) to be prepared in the next CA procedure based on the access status in the previous CA procedure, and grants permission to transmit the CA frame. Advertising via SI (status indication) signal,
(3) A CPE having a MAC frame to be transmitted randomly selects one of a plurality of advertised CA slots and transmits a CA frame.
(4) The SCH confirms that the uplink signal (carrier) has disappeared in order for the CPE that has succeeded in the CA procedure, and then sends an SI signal to instruct the transmission of the MAC frame.
(5) When transmission of all MAC frames is completed, the SCH returns to (2) and starts the next CA procedure.
(6) The CPE that has failed in the CA procedure performs a square backoff process (Non-Patent Document 5) in units of CA slots, and then tries the CA procedure again.
The point is that.

Regarding the item (1), the downlink propagation delay time Td _i , the uplink propagation delay time Tu _i and the transmission timing adjustment time Δt _i shown in FIG. 1 are controlled using a control window (for example, a period of 5 msec) different from the CA procedure. SCH is obtained by sequentially specifying a CPE, returning a response signal, and measuring the round-trip propagation delay time and signal level between the SCH and the CPE, and serves as a base for access synchronization. As a result, when multiple CA frames are transmitted to the same CA slot, the collision is surely detected by entering the collision state from the head of them, or the MAC frame transmitted by the CPE that has succeeded in the CA procedure is changed to another frame. It is possible to adjust the transmission timing so as not to collide with the signal. This means that the restrictions on the network length and frame length that are unavoidable with Ethernet and the like can be eliminated. Refer to Non-Patent Document 7 for details.

  Items (2) to (6) relate to the CA procedure. The SI signal contains the number of CA slots, control information for instructing the transmission of CA frames and MAC frames, and the CA frame contains random numbers for collision detection. Data, error check code, MAC frame length, etc. are described. The SCH checks the bit error of the received CA frame, and if there is no error, the collision does not occur, that is, the CA procedure is instructed to instruct the CPE to transmit the MAC frame using the SI signal in order. .

  In the CSMA / CA method used in wireless LAN, the CA procedure is led by the terminal, and only one MAC frame can be transmitted in one CA procedure. In contrast, in the S-CSMA / v-MCA scheme, the SCH variably controls the number of CA slots according to the load (new generation and retransmission after backoff processing) traffic using the algorithm described in Non-Patent Document 9. By doing so, the probability of collision in the CA procedure is reduced. As a result, the number of CA slots is small and the CA procedure time is shortened in a low load state, so it exhibits a low delay time characteristic. Conversely, in a high load state, a large number of CA slots are prepared and the CA procedure has a high probability of success. Show properties.

  However, in this method, a maximum round trip time (RTT) that depends on the maximum network length is present for each CA procedure or MAC frame transmission instruction. For this reason, as the network length becomes longer or the physical transmission speed becomes faster, the degradation of the throughput characteristics due to the intervention of two RTTs (time waste) becomes more conspicuous. It has also been pointed out that in order to improve the performance of this method, it is necessary to solve problems such as advancement of transmission timing (Non-Patent Document 9).

[Flying S-CSMA / v-MCA method]
Flying S-CSMA / v-MCA method, instead of (4) above,
(4-1) The SCH predicts the time when the upstream frame arrives at the SCH based on the MAC frame length and RTT reported to the CPE during the CA procedure (hereinafter referred to as carrier disappearance time). ,
(4-2) Instruct the CPE to send the frame in advance at the same time.
This is the point.

  This means that the influence of the RTT is reduced as the load increases by causing the CA frame and the MAC frame whose transmission order is second or later (hereinafter referred to as 2nd MAC) to be transmitted in advance in accordance with the carrier disappearance time. As a result of evaluation by simulation, a dramatic improvement in throughput in the long network area (0.73 compared to the maximum throughput of 0.36 of the S-CSMA / v-MCA method at a network length of 40 km) was shown. However, it has also been confirmed that the network length increases slightly as the network length increases (non-patent document 10).

  To further improve performance, the first MAC frame after the CA procedure (hereinafter 1st MAC) must be able to be transmitted in advance. For this purpose, the improvement in the item (5), that is, not “start the next CA procedure after all MAC frames have been transmitted”, It is necessary to be able to start the CA procedure. This is a problem caused by a so-called in-band signaling (hereinafter abbreviated as IBS) method in which access control and payload transmission use the same band.

The analog telephone network used to use the IBS system, but the digital telephone network (ISDN, etc.) introduced an out-of band signaling (hereinafter abbreviated as OBS) system that can provide a variety of call control and services. Became. If we introduce the OBS method to this system based on this knowledge,
Access control and MAC frame transmission can be performed independently, and not only the improvement in throughput characteristics due to the improvement in (5), but also the delay time characteristics, which is ideally independent of the network length and physical transmission rate. It is considered that a good performance can be obtained. It is clear from the example of the digital telephone network that this is also effective in enhancing the expandability of the system such as introduction of communication quality control.

  However, with the introduction of the OBS method, for example, if the CA procedure is continued while ignoring the transmission status of the MAC frame, the MAC frame that has been successful in the CA procedure but is waiting for transmission may become stuck depending on the load status of the network. There is a risk that the delay characteristics will deteriorate rapidly. Some flow control needs to be applied to the CA procedure. It will be described that these problems are solved by the full-flying S-CSMA / v-MCA method incorporating the OBS method and flow control according to the present embodiment.

  Hereinafter, the full-flying S-CSMA / v-MCA scheme according to the present embodiment will be described.

  Note that the present embodiment is not limited to a wireless network, and can be widely applied to a network that is a combination of wireless and a wired transmission medium, or a network that includes only a wired transmission medium.

[Full-flying S-CSMA / v-MCA method]
・ Outline of the method
The full-flying S-CSMA / v-MCA method
Instead of (5)
(5-1) The CA frame transmission instruction for the CA procedure and the MAC frame transmission instruction after the CA procedure are made independent,
(5-2) The ratio of the transmission capacity of the CA channel (uplink for CA frame transmission, uplink control channel) and the MAC channel (uplink for MAC frame transmission, data channel) is appropriate under high load conditions. To further distribute
(5-3) If the MAC procedure that was successful in the CA procedure, but could not be sent and is waiting for a while, is temporarily accumulated, delay the start of the CA procedure.
This is the point.

  In the paragraph (5-1), the introduction of the OBS method starts the next CA procedure without waiting for all CPEs that have succeeded in the CA procedure to finish sending MAC frames, and also starts and ends the CA procedure. It means that the transmission of the MAC frame can proceed without depending on.

  Item (5-2) allocates the transmission capacity corresponding to the number of CA slots to the MAC channel in a high load state where the upper layer generates payloads one after another (for example, normalized load traffic ρ = 1 to 2). In this way, the bandwidth utilization efficiency of the network is to be increased as much as possible. Here, the normalized load traffic is obtained by normalizing the load traffic with the physical transmission rate (the sum of the CA channel and the MAC channel).

  In (5-3), for example, when long MAC frames are concentrated, MAC frames waiting to be transmitted may temporarily accumulate, and the delay time may increase rapidly. It functions as a flow control that suppresses further accumulation of MAC frames.

  The OBS method is a physical OBS method that physically separates channels by frequency division, etc., and a physical IBS / logical OBS method that physically separates channels by time division while sharing the same bandwidth physically. (Non-patent Document 12). Although both implementation methods differ greatly, there is no significant difference in performance if the CA procedure and the MAC frame transmission instruction can proceed independently.

  First, determine the appropriate transmission capacity distribution ratio as a general solution without limiting the implementation method, and the theoretical performance limit derived from this, then limit the target model and the physical implications of bandwidth allocation and channel occupancy, Furthermore, the application method of flow control to CA procedure is considered.

(I) Proper transmission capacity distribution ratio
If the number of CA slots prepared by the SCH for each CA procedure is n, and the number of CPEs (simultaneous access) to be performed by selecting one of these at random is k, the CA frame is received. The average number of CA slots that did not exist A _N , the average number of CA slots that received only one CA frame (CA procedure succeeded) A _S , the average number of CA slots that received two or more CA frames (CA procedure failed) A _C each

It is represented by

  Here, when a plurality of CPEs waiting for the CA procedure proceed to the CA procedure all at once, a collision between CA frames always occurs and the procedure fails when n = 1. In order to avoid this, in this embodiment, n> = 2 is set. Note that the wireless LAN CSMA / CA method corresponds to n = 1, but even if carrier sensing is performed to recognize an idle state, collisions can be avoided by waiting for a random time before transmitting a frame. ing.

As k increases, A _N decreases and, conversely, A _C increases, but A _S has a peak value that depends on k and n. This average peak value A _Smax is

And has a peak value when k is one less than or equal to n. At that time, the average CA procedure success probability P _S per CA slot is also the peak value.

have. This, k is P _S is maximized when a one less or equal than n, i.e. be provided CA slot of the same number as the k expected, at least an average 36.8% over the CPE CA procedures (Non-patent Documents 8 and 9).

Also, assuming that the length per CA slot (≈CA frame length) is FL _CA and the average length of MAC frames is FL _MAC , the transmission capacity CP _CA of the CA channel required to accommodate n CA slots is: Product of occupied time TP _CA and transmission band BW _CA of the same channel

It is represented by On the other hand, the transmission capacity CP _MAC of the MAC channel required to transmit all MAC frames expected to be successful in the CA procedure is the product of the occupied time TP _MAC and the transmission bandwidth BW _MAC.

It is represented by Therefore, the proper transmission capacity distribution ratio ^ m described in (5-2) is

Can be expressed as

At this time, the maximum throughput ^ φmax obtained is

It becomes. Since this equation is represented only by ^ m determined by FL _MAC , FL _CA, and P _Smax , it can be seen that ^ φ does not depend on the network length or physical transmission rate.

By the way, as shown in Table 1 below, FL _CA = 64bits, FL _MAC = 2838 bits (average MAC frame length when the ratio of short frames to long frames is 8: 2), and CA procedure is repeatedly executed Therefore, assuming that infinite number of CA slots are randomly accessed and P _Smax = 0: 368, ^ m = 0: 0613 is obtained. ^ Φmax at this time is 0.942 from the equation (9), which represents the theoretical performance limit of the method of the present embodiment.

(II) Limitation of target models In order to make the following discussion easier to understand, the target models are limited as follows. In other words, an implementation model is considered in which a physical OBS scheme that frequency-divides a CA channel and a MAC channel using OFDM modulation or the like is employed, and TP _CA ≈TP _MAC . Therefore, in this implementation model, the appropriate transmission capacity allocation ratio ^ m in equation (8) is the bandwidth allocation ratio between BW _CA and BW _MAC.

It is represented by

Figure 2 assumes a high load condition (ρ = 1 to 2) that consumes almost 100% of the network bandwidth.
(a) TP _CA > TP _MAC
(b) TP _CA <TP _MAC
(c) TP _CA ≒ TP _MAC
The channel occupancy state in the SCH is shown for three cases in which the bandwidth allocation ratio between the BW _CA and the BW _MAC is set so that _TRTT in the figure represents the round-trip propagation delay time from when the SCH instructs transmission to the arrival of the frame at the SCH, including the processing time of the SCH and CPE and the adjustment time of transmission timing. include.

FIG. 4A shows a case where BW _CA is set too small for an appropriate distribution ratio, and TP _CA > TP _MAC . Even if all the MAC frames that have passed the q-th CA procedure have been sent, the (q + 1) -th CA procedure has not been completed, and (q + 1) ) Transmission of the second MAC frame is started. This means that the MAC channel is not 100% used, that is, the bandwidth allocation of the network is not appropriate.

On the other hand, FIG. 5B shows a case where BW _CA is set excessively and TP _CA <TP _MAC . Since the time required for the (q + 1) -th CA procedure is shorter than the time required for transmitting all the MAC frames that were successfully passed the q-th CA procedure, all the MAC frames that were successfully passed the q-th CA procedure were sent. (Q + 1) -th and (q + 2) -th CA procedures start one after another. The time difference ΔTP _q between TP _MAC and TP _CA generated for this is
Δ TP _q = TP _MACq -TP _CAq (11)
It is represented by However, _{ΔTP q} = 0 when TP _MACq ≦ TP _CAq .

  In other words, although the CA procedure is successful, MAC frames that cannot be transmitted and are kept waiting are accumulated one after another. If this state persists, the excess capacity of the network will be lost, and the meaning of decentralized access by the square back-off process will be lost, that is, the basic operation principle of the competitive access method will be lost.

On the other hand, in the case where the BW _CA of Fig. (C) is set appropriately and the average TP _MAC and TP _CA are almost equal, the time taken to transmit all the MAC frames that have succeeded in the qth CA procedure , (Q + 1) time required for the CA procedure is almost equal. There is no wasted time gap between CA procedures and between sending MAC frames, and there is no loss of excess network capacity. For this reason, there is no sudden increase in delay time, and the utilization efficiency of each channel is the highest. That is, in the implementation model, it is understood that the distribution ratio ^ m between BW _CA and BW _MAC may be set so that TP _CA and TP _MAC are substantially equal.

(III) Flow control for CA procedure Even if ^ m is set appropriately based on the previous section, it cannot be denied that in a real environment where traffic fluctuates from moment to moment, the delay time increases temporarily. The flow control for the CA procedure is intended to avoid such a situation, and when the state of TP _MAC > TP _CA continues as shown in Fig. 3, the CA procedure is equivalent to the time difference ΔTP of equation (11). If the start of the delay is delayed, accumulation of delay time is suppressed.

However, activation of flow control results in a reduction in throughput because the CA procedure is thinned out. In addition, the delay time fluctuates as the CA procedure interval fluctuates. The latter is dominated by MAC frame congestion and is negligible. On the other hand, even if the former may momentarily become TP _MAC > TP _CA , if the preceding and following CA procedures are TP _MAC << TP _CA , the flow control is not activated, so that throughput in a low load state is reduced. Although a slight decrease can be suppressed (in the simulation described later, about 1% at around ρ = 0: 3), another measure is required in a high load state. In other words, it is a problem to satisfy the contradictory performance indexes of suppressing the delay time and improving the throughput in a balanced manner. The introduction of a flow control suppression coefficient, which will be described later, will provide a solution.

In the above discussion, several assumptions were made, such as ρ = 1 to 2, short frame number ratio 8: 2, P _Smax = 0: 368, but it is rare to satisfy these in the actual environment. is there. Therefore, ^ m, ΔTP, and the like should be taken as providing guidelines (design criteria) for the implementation of the method, and some of the coefficients including the above coefficients will be introduced and consideration will be given later.

[Specification of system and evaluation of effectiveness]
-Specification of method FIG. 4 shows a specific example of the mounting model and its operation mechanism. In this figure, in order to avoid complication, the transmission interval of the SI signal is expanded from Table 1 described later. The specific operation is as follows.

1) When the SCH sends (advertises) the SI signal declaring the start of the q-th CA procedure (number of CA slots n) before T _RTT from the (q-1) -th scheduled CA slot end time, the signal is to arrive at the CPE _i after Td _i.

2) The CPE _i having the MAC frame to be transmitted randomly selects one of the n CA slots, and after a predetermined transmission timing Δt _i = T _RTT- (Td _i + Tu _i ) When transmitting on the CA channel, it arrives at the selected CA slot position (the first slot in the example of FIG. 4) after Tu _i . The same applies to other CPEs.

  3) The SCH checks the bit error of the received CA frame and determines the success or failure of the CA procedure.

4) The SCH uses the SI signal to advertise the access result to the CPE as bitmap information (“1001” in the example of FIG. 4). However, if a carrier remains on the MAC channel, the advertisement is postponed until _TRTT before the predicted disappearance time of the carrier.

5) Receiving this, CPE _i knows that the CA procedure was successful, and starts sending MAC frames after Δt _i .

6) SCH, based on the MAC frame length notified in CA frame from the CPE _i, predicts a carrier annihilation time on MAC channel, the T _RTT before the same time, updated bitmap information (in FIG. 4 In the example, “0001”) is advertised. Thereafter, the operations in 5) to 6) are repeated until transmission of all the MAC frames that have been successfully performed in the CA procedure is completed.

7) On the other hand, CPEj and CPE (not shown) that sent the CA frame to the slot position that is “0” in the bitmap information (the third slot in the example of FIG. 4 ) failed in the CA procedure. Knowing that, we move on to the back-off process and prepare for the next CA procedure.

8) The operations in 1) to 7) are repeatedly executed, but when the flow control for the CA procedure is performed, the start of the (q + 2) th CA procedure is declared with a delay of ΔTP _q .

  The items 1) to 3) and 4) to 6) are the independent progress of the CA procedure and the MAC frame transmission instruction, and the item 8) is the flow control. Further, the preceding transmission of the CA frame is executed by the item 1), the first MAC is the item 4), and the transmission after the 2nd MAC is the item 6).

  FIG. 5 is a diagram showing a configuration of the PTMP access network according to the present embodiment.

  As shown in the figure, in the PTMP type access network according to the embodiment, user computers 1 (1-1 to 1-6) are connected to CPE 3 (3 through router 2 (2-1 to 2-3). -1 to 3-3). Each CPE 3 (3-1 to 3-3) is connected to the SCH 4 as a base station via a wireless network, and is connected to the Internet / router 5.

  FIG. 6 is a diagram showing a SCH configuration of the PTMP access network according to the present embodiment.

  In the figure, an external interface 11 is an interface for exchanging data with the Internet / router 5. The IP packet transmission buffer 12 is a buffer that temporarily stores IP packets input from the Internet / router 5 via the external interface 11 and transmitted to the CPE 3. The MAC frame generator 13 generates and adds a MAC frame to the IP packet output from the IP packet transmission buffer 12 and outputs the MAC frame to the modulator 14 according to the transmission timing control signal from the transmission timing control circuit 18. The modulator 14 outputs the SI signal generated by the SI signal generator 17 and the MAC frame generated by the MAC frame generator 13 from the transmission timing control circuit 18 using the carrier / clock from the carrier / clock generator 19. The signal is modulated according to the transmission timing control signal and output to the frequency converter 15. The frequency converter 15 up-converts the modulation signal output from the modulator 14 with the carrier / clock from the carrier / clock generator 19 and outputs the result to the transmission power amplifier 16. The transmission power amplifier 16 amplifies the power of the up-converted modulated signal from the frequency converter 15 and outputs it to the antenna 21 via the branching unit 20. Thereby, a signal (data) is transmitted from the SCH 4 to the CPE 3 via the wireless network.

  A frame such as a CA frame and a MAC frame transmitted from the CPE 3 and received by the antenna 21 is input to the reception amplifier 22 via the branching unit 20. The reception amplifier 22 amplifies the received frame and outputs it to the frequency converter 23. The frequency converter 23 down-converts the frame output from the reception amplifier 22 with the carrier / clock from the carrier / clock generator 19 and outputs the result to the demodulator 24. The demodulator 24 demodulates the down-converted frame from the frequency converter 23 according to the carrier / clock from the carrier / clock generator 19 and outputs it to the frame type discriminator 25. The frame type discriminator 25 discriminates the type of the down-converted frame from the demodulator 24. The CA frame analyzer 26 analyzes the CA frame discriminated by the frame type discriminator 25 and outputs an SI signal generation instruction to the SI signal generator 17 based on the analysis result. The SI signal generator 17 outputs the SI signal to the modulator 14 in accordance with the SI signal generation instruction from the CA frame analyzer 26 and the transmission timing control signal from the transmission timing control circuit 18. The MAC frame decomposer 27 decomposes the MAC frame determined by the frame type determiner 25 and outputs the IP packet to the IP packet reception buffer 28. The IP packet reception buffer 28 temporarily stores the IP packet from the MAC frame decomposer 27 transmitted to the Internet / router 5 via the external interface 11.

  FIG. 7 is a diagram showing a CPE configuration of the PTMP access network according to the present embodiment.

  In the figure, an external interface 31 is an interface for exchanging data with a user's computer 1. The IP packet transmission buffer 32 is a buffer that temporarily stores IP packets input from the computer 1 via the external interface 31 and transmitted to the SCH 4. The MAC frame generator 33 generates and adds a MAC frame to the IP packet output from the IP packet transmission buffer 32, and outputs the MAC frame to the modulator 34 according to the transmission timing control signal from the transmission timing control circuit 38. The modulator 34 transmits the CA frame generated by the CA frame generator 37 and the MAC frame generated by the MAC frame generator 33 from the transmission timing control circuit 38 using the carrier / clock from the carrier / clock synchronous regenerator 39. And is output to the frequency converter 35. The frequency converter 35 up-converts the modulated signal output from the modulator 34 with the carrier / clock from the carrier / clock synchronous regenerator 39 and outputs the result to the transmission power amplifier 36. The transmission power amplifier 36 amplifies the power of the up-converted modulation signal from the frequency converter 35 and outputs it to the antenna 41 via the branching unit 20. Thereby, a signal (data) is transmitted from the CPE 3 to the SCH 4 via the wireless network.

  Frames such as CA frames and MAC frames transmitted from the SCH 4 and received by the antenna 41 are input to the reception amplifier 42 via the branching unit 40. The reception amplifier 42 amplifies the received frame and outputs it to the frequency converter 43 and the carrier / clock synchronous regenerator 39. The frequency converter 43 down-converts the frame output from the reception amplifier 42 with the carrier / clock from the carrier / clock synchronous regenerator 39 and outputs the result to the demodulator 44. The demodulator 44 demodulates the down-converted frame from the frequency converter 43 according to the carrier / clock from the carrier / clock synchronous regenerator 19 and outputs it to the frame type discriminator 45. The frame type discriminator 45 discriminates the type of the down-converted frame from the demodulator 44. The SI signal analyzer 46 analyzes the SI signal discriminated by the frame type discriminator 25 and outputs a CA frame generation instruction to the CA frame generator 37 based on the result of the analysis. The CA frame generator 37 outputs the CA frame to the modulator 34 in accordance with the CA frame generation instruction from the SI signal analyzer 46 and the transmission timing control signal from the transmission timing control circuit 38. The MAC frame decomposer 47 decomposes the MAC frame determined by the frame type determiner 45 and outputs the IP packet to the IP packet reception buffer 48. The IP packet reception buffer 48 temporarily stores the IP packet from the MAC frame decomposer 47 transmitted to the computer 1 via the external interface 31.

  FIG. 8 is a diagram illustrating SCH state transition.

  As shown in the figure, the SCH in the initial state J1 calculates the transmission timing from the uplink MAC frame disappearance time, transmits the transmission permission to the CPE via the SI signal at the calculated transmission timing, and enters the state J2. Transition. In the state J2, the reception of the CA frame from the CPE is waited during the RTT, and when the CA frame is not received during the RTT, the state returns to the state J1.

  On the other hand, when the SCH receives the CA frame, the state transits to the state J3, and enters the state J3 waiting for CA transmission permission. State J3 is a state of waiting for the downstream Si timing or flow control.

When the transmission timing (Si timing) is reached, a CA transmission permission is transmitted to each CPE _i, and the state transits to a state J4 waiting for MAC frame reception. On the other hand, if the CA transmission permission is not the predetermined time, the process returns to the state J2.

  The state J4 is a state of waiting for reception of a MAC frame. When the MAC frame cannot be received for a predetermined time, the state returns to state J3, and when the MAC frame is received from any CPE, the state transitions to state J5.

  The state J5 is a state of waiting for the completion of the reception of the MAC frame. When the reception of the MAC frame is finished and there is a next CPE, the state J3 is entered, the reception of the MAC frame is finished, and the next If there is no CPE, the state transitions to state J1.

  FIG. 9 is a diagram illustrating state transition of CPE.

  As shown in the figure, the CPE in the initial state J11 transitions to the state J12 when there is a transmission MAC frame. The state J12 is a state of waiting for reception of CA transmission permission. When the CA transmission permission is received, the state J12 transits to state J13. On the other hand, if the CA transmission permission is not received even after the RTT has elapsed, the process transits to the initial state J11.

In the state J13, after the transmission timing adjustment time Δt _i , the CA frame is transmitted to the SCH, and the state transitions to the state J14. The state J14 is a waiting state for MAC frame transmission permission reception, and when the MAC frame transmission permission is granted, the state J14 is shifted to the state J16. On the other hand, when the MAC frame transmission cannot be permitted, the flow goes to the state J13. Furthermore, when the MAC frame transmission permission fails, that is, when the MAC frame transmission collides, the state transitions to state J15.

  The state J15 is a state waiting for transmission back-off, and after the transmission back-off process, transits to state J12. The state J16 is a state waiting for the end of MAC frame transmission. After waiting for the end of transmission of the MAC frame, the state J16 transitions to a state J17 during MAC frame transmission.

  10 and 11 are flowcharts for explaining the operation of the SCH.

  First, the SCH determines whether there is a value of the MAC reception disappearance time t1 (S1). If it is determined in S1 that there is no value for t1, t1 = 0 is set (S2), and the process proceeds to S3. If it is determined in S1 that there is a value of t1, the process proceeds to S3.

  In S3, it is determined whether or not the current time t is t> t1 (S3). If it is determined in S3 that the current time t is t> t1, if the transmission capacity (bandwidth allocation) between the CA channel and the MAC channel is changed in S11 described later, the changed transmission is performed. The capacity is advertised to each CPE (S4). Note that the transmission capacity notification in S4 may be performed before the MAC transmission permission in S19. Next, a CA transmission permission is transmitted (S5), the timer T_rtt = RTT time is set (S6), and it is determined whether or not a CA frame has been received (S7).

  If it is determined in S7 that a CA frame has not been received, the timer T_rtt is subtracted by the elapsed time (S8), and it is determined whether or not the timer T_rtt ≦ 0 (S9). If it is determined that the timer T_rtt ≦ 0 is not satisfied, the process returns to S7. If it is determined that the timer T_rtt ≦ 0, it is determined whether or not the operation is to be ended (S10). If it is determined that the operation is not to be ended, the processing returns to S1 and the operation is ended. If it is determined, the operation is terminated.

  On the other hand, if it is determined in S7 that a CA frame has been received, the bandwidth load is calculated from the access state for each slot, and if the calculated bandwidth load reaches a predetermined load, the transmission capacity (Bandwidth allocation) is changed (S11). Next, it is determined whether there is a CPE that has been successfully accessed (S12). If it is determined that there is no CPE that has been successfully accessed, the process returns to S5. On the other hand, if it is determined that there is a CPE that has been successfully accessed, a flow control calculation (confirmation of the MAC accumulated state) is performed (S13), and whether or not the flow control is necessary based on the accumulated state. A determination is made (S14).

  If it is determined in S14 that flow control is necessary, a standby time is calculated (S15), a standby time timer is started (S16), and it is determined whether or not the standby time has ended (S16). S17).

10 and 11 , the flow control S15 to S17 is executed in the SCH. However, the flow control may be executed in the CPE. In this case, the flow control calculation in S13 is executed in the SCH, and it is determined whether or not the flow control is necessary in S14. If it is determined that the flow control is necessary, the flow control is necessary. Notify each CPE that there is. Each CPE executes the flow control of S15 to S17 when advertised from the SCH that the flow control is necessary.

If it is determined in S17 that the waiting time has expired or if it is determined in S14 that flow control is not required, CPE _i is selected (S18).

Next, a MAC transmission permission is transmitted to the selected CPE _i (S19), timer Tmac = T_MACi is set (S20), subtraction of the timer Tmac is started (S21), and whether Tmac> 0 is satisfied. Is determined (S22). If it is determined that Tmac> 0 is not satisfied, it is determined whether or not there is a remaining MAC transmission waiting CPE _i (S23), and if it is determined that there is a remaining MAC transmission waiting CPE _i. , The process returns to S18.

On the other hand, if it is determined that there is no remaining MAC transmission waiting CPE _i , the MAC reception disappearance time t1 is calculated (S24), and it is determined whether or not t1−current time t−T_rtt ≦ 0. (S25). If it is determined in S25 that t1−current time t−T_rtt ≦ 0 is not satisfied, the process returns to S24. If it is determined that t1−current time t−T_rtt ≦ 0, the process returns to S1. .

  FIG. 12 is a flowchart for explaining the operation of the CPE.

  First, the CPE determines whether there is a transmission MAC frame (S31). In S31, when it is determined that there is no transmission MAC frame, it is determined whether or not the communication is terminated (S32). When it is determined that the communication process is terminated, the process is terminated. On the other hand, if it is determined that the communication has not ended, the process returns to S31.

If it is determined in S31 that there is a transmission MAC frame, it is next determined whether or not CA transmission permission has been received from the SCH (S33). When it is determined that the CA transmission permission has been received, if the change of transmission capacity (bandwidth allocation) is advertised from the SCH, the transmission bit rates of the CA channel and the MAC channel based on the advertised transmission capacity (S34), wait for the transmission timing adjustment time Δt _i (S35), and transmit the CA frame (S36).

  Next, it is determined whether or not the MAC frame transmission permission has been received (S37). If it is determined that the MAC frame transmission permission has been received, it is determined whether or not the MAC frame can be transmitted (S38). If it is determined that the MAC frame cannot be transmitted, a back-off process is performed (S39), and the process returns to S33.

  On the other hand, when it is determined in S38 that the MAC frame can be transmitted, it is determined whether or not the transmission of the MAC frame is permitted (S40). Returning to the process, if it is determined that the transmission is permitted by itself, Δti is waited (S41), the MAC frame is transmitted (S42), and the process returns to S31.

  In this embodiment, the operation having the means for changing the bandwidth capacity in addition to the flow control has been described. However, a system without means for changing the bandwidth capacity may be used. Even in this case, the throughput can be improved by the flow control. Specifically, the processing in S4 and S11 of the flowchart of SCH shown in FIG. 10 and S34 shown in FIG. 12 may be omitted.

-Characteristic evaluation by simulation Generally, the superiority or inferiority of the access control method is evaluated using throughput and delay time as objective functions and load traffic, network length, average frame length, etc. as variables. From this same viewpoint, in order to evaluate the effectiveness of the proposed method, the network environment conditions covered by this system shown in Table 1 (the target area of the uplink physical transmission rate is 5 to 20 Mbps, but the traffic If the bandwidth in the downstream direction with a large amount of data is widened (for example, 4 times that of upstream), and a limited radio band is repeatedly used with a sector antenna (for example, 15 sectors, 3 frequency repetition), a gigabit-class access network The simulation was performed. N = 32 in the table is an appropriate value of the upper limit number in the above-described variable control of the number of CA slots. Refer to References 9 and 10 for details.

  In the simulation, one SCH using 1 bit time (unit converted to transmission time per bit on the MAC channel) on the computer and a CPE capable of operating up to 5000 units at the same time were constructed. 4 and 1) to 8) are implemented so that the state transitions in units of SI signal transmission interval (256bit-times). In addition, an IP packet arrives at Poisson from the upper layer at a short MAC frame ratio of 8: 2, which is one of the traffic models based on actual measurements at our campus, and the MAC frame transmission of the packet is completed. It is assumed that no new packet arrives (Non-Patent Document 11).

  After starting simulation under certain applied (newly generated) traffic conditions, it takes about 2: 5 x 107bit-times (about 2.5 seconds if the physical transmission speed is 10Mbps) until the load traffic becomes steady. The period from 3 × 107 to 5: 5 × 107 bit-times was the observation window period (Δw), and the average value was calculated. Furthermore, in order to prevent bias in the simulation results due to applied traffic, the simulation was performed five times with different initial values to obtain a total of five average values, and the average value was used as observation data. Incidentally, in the following simulation results, the standard deviation of the five throughput average values around ρ = 1 was 2% or less of the average value. Also, the sum of the time taken from the new generation of a MAC frame successfully transmitted during Δw to the completion of transmission and the time taken from the new generation to discard of a MAC frame that failed to be retransmitted during Δw. Is divided by the number of MAC frames successfully transmitted during Δw as the delay time. The standard deviation of the average value of the five delay times near ρ = 1 obtained in the same manner as the throughput was 2% or less of the average value.

  The simulation program is collated with theoretical calculation values at several singular points such as ρ << 1, ρ = 1, maximum throughput, short length MAC frame ratio 0:10, and correspondence between simulation results and implemented operations We verified the legitimacy of the operation through attaching.

In the following description, the SCSMA / v-MCA method that does not apply the preceding transmission is the NF method, the flying S-CSMA / v-MCA method that performs the preceding transmission after the CA frame and 2nd MAC is the F method, and the 1st MAC is also advanced. The full-flying S-CSMA / v-MCA method to be transmitted is abbreviated as FF method.

(I) Validity of bandwidth allocation ratio ^ m
In order to evaluate the validity of ^ m shown in Eq. (10), ^ m was multiplied by the CA band expansion coefficient β and a simulation was performed with the reference values in Table 1 (fixed to 32 slots). Here, β> 1 means that BW _CA is expanded β times and TP _CA is shortened, that is, TP _CA <TP _MAC in FIG.

FIG. 13 shows changes in characteristics due to β. Incidentally, <when the _{1 (BW MAC × ρ) =} (BW MAC + BW CA), ρ> the theoretical limit value figures [rho was calculated as = 1 BW _MAC when the = (BW _MAC + BW _CA) Is.

(A) Throughput characteristics It can be seen from the throughput characteristics shown in FIG. 6A that the larger β is, the closer to 90% of the throughput is exhibited up to the overload region (ρ> 2). The maximum throughput close to 90% is close to the theoretical performance limit ^ φ = 0: 942 obtained from equation (9), and it is confirmed that the FF method has high potential.

(B) Delay characteristics In the delay characteristics of (b) in the figure, the larger the β, the greater the delay time from around ρ = 1, but this is due to the stagnant MAC frame and the square backoff. You can see that the process doesn't make sense. On the contrary, as β approaches 1, it can be seen that although the high throughput cannot be obtained, the delay time increases slowly.

In the above example, the ratio of short MAC frames was set to 8: 2. However, for 10: 0 and 0:10, ^ m was calculated using equation (10) and a simulation was performed. It was. From these, it is considered significant to set the distribution ratio of BW _CA and BW _MAC so that TP _CA and TP _MAC are substantially equal.

(II) Effectiveness of flow control
In order to evaluate the effectiveness of flow control for the CA procedure, ΔTP _q = TP _MACq -γ × TP _CAq (12) with the flow control suppression coefficient γ added to the above equation (11)
The simulation was performed under the same conditions as in (I) with β = 1: 7 as an example. However, when TP _MACq ≦ TP _CAq , ΔTP _q = 0. Here, γ = ∞ means that no flow control is performed, and γ = 1 means that flow control is performed based on the equation (11). Note that flow control is applied when TP _MAC > TP _CA continuously for two or more times.

FIG. 14 shows a change in characteristics due to γ. As γ is increased, better throughput characteristics are obtained, but flow control becomes difficult to start. It can be seen that the delay time increases rapidly. On the contrary, the throughput is suppressed as γ approaches 1, but the rapid increase in delay time is also suppressed, and it can be seen that the flow control functions steadily. Note that when γ = 1: 3, a sudden increase in delay time is suppressed when an overload condition of 4 <ρ is reached. This is because delay time due to MAC frame congestion is dominant when 1 <ρ <4, whereas delay time due to retransmission or discard due to CA procedure failure is dominant when 4 <ρ.

  From the above, it can be seen that when β = 1: 7, there is an appropriate value in the vicinity of γ = 1: 2, where the suppression of delay time and the improvement of throughput are well balanced.

(III) Comparison of characteristics with the conventional method In the conventional NF method and F method, the longer the network length and the faster the physical transmission speed, the more the throughput characteristics deteriorate, but the FF method is not affected by these. Realization of special characteristics is expected. In order to confirm this, the characteristics of these three methods (all of which the number of CA slots is variable) were compared in the application area centering on the reference values in Table 1.

  In the CA slot variable control, the number of CA slots n to be prepared in the next CA procedure is determined based on the access status in the previous CA procedure and the time interval of the CA procedure. The prediction upper limit risk factor α used in this is a coefficient for avoiding a shortage of n due to an underestimation of the number of simultaneous accesses. The NF method and F method that sandwich the transmission of MAC frames between CA procedures greatly change the CA procedure interval and easily cause excess or deficiency of n. The FF method with a small width is set to α = 1 (see Non-Patent Document 9).

  Further, β = 1: 7 and γ = 1: 2 based on the evaluation results of (I) and (II).

  FIG. 15 shows how n changes with respect to ρ. In the NF method and the F method, 10 or more CA slots are already prepared near ρ = 1, while in the FF method, ρ = 2. It can be seen that the increase in n is suppressed to the vicinity. This reflects the difference in α, but the root is due to the difference in method between the IBS method and the OBS method.

  Figure 16 shows the characteristics of several combinations of network length (2.5, 10, 40 km) and physical transmission speed (5, 10, 20 Mbps) with the same ratio of short MAC frames as the previous example. This is an example of comparison. In addition, the figure also shows the characteristics of the 802.11a wireless LAN.

(A) Throughput characteristics In the throughput characteristics shown in (a) of the figure, it is confirmed that the five line graphs for the FF method overlap and show almost the same characteristics without depending on the network length or physical transmission rate. It can be seen that the FF method can obtain a throughput of 12 to 21% higher than the F method (2.5 to 40 km, 10 Mbps) and 38 to 142% higher than the NF method (10 to 40 km, 10 Mbps).

(B) Delay characteristics Regarding the delay characteristics shown in (b) of the figure, the FF method shows almost the same characteristics in the FF method, although there is a slight increase in the delay time due to MAC frame congestion around ρ = 2-4. ing. And for example, around ρ = 0: 6 (10km, 10Mbps), the delay time of F method and NF method is 1.5msec and 2.0msec, respectively, but the FF method is 0.9msec, decreasing to around half, which is good It can be seen that the delay time characteristic is excellent.

  From the above, it is confirmed that the FF method exhibits high throughput characteristics independent of network length and physical transmission speed while maintaining low delay characteristics.

  Note that the maximum throughput in the vicinity of ρ = 3 in the FF method is about 6% deviation from the theoretical performance limit ^ φ, but there is about 20% deviation in the vicinity of ρ = 1. This divergence seems to be able to be improved somewhat by closely examining α, β, and γ, but by securing the surplus capacity of the network by square backoff processing, it tries to respond flexibly to a large number of burst transfer requests This is thought to be the cause.

  According to the present embodiment, the fullflying S-CSMA / v-MCA scheme is intended to advance the transmission efficiency of the upstream transmission medium as much as possible by preliminarily transmitting all CA frames and MAC frames for the carrier disappearance time on the upstream channel. Developed. One of the points in realizing the system is that the CA procedure and the MAC frame transmission instruction can proceed independently by introducing the OBS system that allows flexible access control. However, under high load conditions, although the CA procedure was successful, the CPE that could not wait for transmission of the MAC frame was accumulated and the delay time increased rapidly. It was shown that this problem can be solved by appropriately setting the CA channel and MAC channel bandwidth allocation ratio ^ m and applying flow control to the CA procedure.

  As a result of evaluating the characteristics by theoretical calculation and simulation, the maximum throughput of nearly 90% is achieved without depending on the network length and physical transmission speed of several tens of kilometers while maintaining the low delay characteristic that is the characteristic of the competitive method. It was confirmed.

  ^ M depends on the ratio of the number of short and long MAC frames flowing on the network. If SCH is grasped one by one, and ^ m is adaptively controlled in combination with OFDM modulation or spread modulation, the network can be maintained in a more appropriate state.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  DESCRIPTION OF SYMBOLS 17 ... SI signal generator, 18 ... Transmission timing control circuit, 25, 45 ... Frame type discriminator, 26 ... CA frame analyzer, 27, 47 ... MAC frame decomposer, 37 ... CA frame generator, 38 ... Transmission timing Control circuit, 46... SI signal analyzer.

Claims (14)

In a hub device in a PTMP network that shares a hub device with a plurality of subscriber devices and performs transmission control between the hub device and the subscriber device,
A different bandwidth is allocated to an uplink control channel for transmitting a CA (collision avoidance) frame between the hub device and the plurality of subscriber devices, and a data channel for transmitting a data frame,
Means for changing transmission capacities of the uplink control channel and the data channel based on a load of the uplink control channel and the data channel when at least one of the uplink control channel and the data channel is in a high load state;
A hub device comprising means for independently performing CA frame transmission control on the uplink control channel and data frame transmission control on the data channel.   The hub apparatus according to claim 1, wherein the means for changing the transmission capacity of the uplink control channel and the data channel changes the transmission capacity to a predetermined ratio between the uplink control channel and the data channel. 3. The hub according to claim 2, wherein the ratio of the transmission capacity is determined based on a length per at least one CA slot accommodating a CA frame in the uplink control channel and an average length of a data frame. apparatus. The hub apparatus according to claim 2, wherein the ratio of the transmission capacities is determined so that an occupation time of the uplink control channel and an occupation time of the data channel are equal. The apparatus further comprises means for delaying transmission of a data frame transmission permission signal to the subscriber unit when a state in which the data channel occupation time is longer than the uplink control channel occupation time continues for a predetermined time or more. The hub device according to Item 1. 6. The hub device according to claim 5, wherein the means for delaying transmission of a data frame to the subscriber device transmits a flow control notification signal. The hub device according to claim 1, wherein the transmission capacity is changed by changing the number of CA slots. In a communication method in a hub device in a PTMP type network in which a hub device is shared by a plurality of subscriber devices and transmission control is performed between the hub device and the subscriber device,
  Assigning different bands to an uplink control channel for transmitting a CA (collision avoidance) frame and a data channel for transmitting a data frame between the hub device and the plurality of subscriber devices;
  When at least one of the uplink control channel and the data channel is in a high load state, the transmission capacity of the uplink control channel and the data channel is changed based on the load of the uplink control channel and the data channel,
  A communication method, wherein CA frame transmission control in the uplink control channel is performed independently of data frame transmission control of the data channel.   In a subscriber device in a PTMP network that shares a hub device with a plurality of subscriber devices and performs transmission control between the hub device and the subscriber device,
  A different bandwidth is allocated to an uplink control channel for transmitting a CA (collision avoidance) frame between the hub device and the plurality of subscriber devices, and a data channel for transmitting a data frame,
  A means for transmitting a CA frame when a CA frame transmission permission signal is received from the hub device;
  When at least one of the uplink control channel and the data channel is in a high load state, the transmission capacity of the uplink control channel and the data channel is changed based on the load of the uplink control channel and the data channel,
  The subscriber apparatus, wherein the CA frame transmission control in the uplink control channel is performed independently of the data frame data frame transmission control.   10. The subscriber unit according to claim 9, further comprising means for delaying transmission of a CA frame when said hub unit is notified that flow control is necessary.   In a communication method in a subscriber apparatus in a PTMP type network in which a hub apparatus is shared by a plurality of subscriber apparatuses and transmission control is performed between the hub apparatus and the subscriber apparatus.
  A bandwidth different from a data channel for transmitting a data frame is allocated to an uplink control channel for transmitting a CA (collision avoidance) frame between the hub device and the plurality of subscriber devices,
  A means for transmitting a CA frame when a CA frame transmission permission signal is received from the hub device;
  When at least one of the uplink control channel and the data channel is in a high load state, the transmission capacity of the uplink control channel and the data channel is changed based on the load of the uplink control channel and the data channel,
  A communication method in which transmission control of a CA frame in the uplink control channel is performed independently of transmission control of a data frame in the data channel.   12. The communication method according to claim 11, further comprising delaying transmission of a CA frame when notified from the hub device that flow control is required.   In a communication system in a PTMP type network in which a hub device is shared by a plurality of subscriber devices and transmission control is performed between the hub device and the subscriber devices.
  A different bandwidth is allocated to an uplink control channel for transmitting a CA (collision avoidance) frame between the hub device and the plurality of subscriber devices, and a data channel for transmitting a data frame,
  The hub device is
  Means for changing transmission capacities of the uplink control channel and the data channel based on a load of the uplink control channel and the data channel when at least one of the uplink control channel and the data channel is in a high load state;
  Means for performing CA frame transmission control in the uplink control channel independently of data frame transmission control of the data channel;
Comprising
  The subscriber unit is
  A communication system comprising means for transmitting a CA frame when a CA frame transmission permission signal is received from the hub device.   In a PTMP type network where a hub device is shared by a plurality of subscriber devices and transmission control is performed between the hub device and the subscriber devices.
Assigning different bands to an uplink control channel for transmitting a CA (collision avoidance) frame and a data channel for transmitting a data frame between the hub device and the plurality of subscriber devices;
  The arrival of the upstream CA frame to the hub device is bit-synchronized at a predetermined timing, and the reception signal level at the hub device is made equal,
  The hub device advertises the time slot for receiving the next CA frame based on the access state of the previous CA frame, and advertises the transmission permission of the CA frame via the downlink status indication signal,
The advertisement is transmitted in advance before the end time of the reception time slot advertised immediately before,
  In response to the advertised status signal indication signal, a subscriber unit having a data frame to be transmitted randomly selects one of the advertised reception time slots and selects the selected reception frame. The subscriber apparatus that transmits a CA frame in a time slot and fails in the collision avoidance procedure performs a predetermined back-off process in units of the reception time slot, and then tries to retransmit the CA frame.
  Based on the data frame length declared by the subscriber device and the maximum round-trip propagation time, the hub device predicts the carrier disappearance time when the upstream data frame arrives at the hub device,
  The hub device instructs the subscriber device to transmit the data frame in advance, aiming at the predicted carrier disappearance time;
  The CA frame transmission permission and the data frame transmission instruction are performed independently in time,
  In the hub device, when at least one of the uplink control channel for CA frame transmission and the uplink data control channel for data frame transmission is in a high load state, based on the load of the uplink control channel and the uplink data channel, Changing the transmission capacity of the uplink control channel and the uplink data channel;
  The subscriber device temporarily delays the start of transmission of the CA frame when data frames that cannot be transmitted and are kept waiting are accumulated;
Multiple access control method.

Images