JPH05236012A - Packet communication system - Google Patents

Abstract

PURPOSE:To give equal communication quality to each terminal station by eliminating the difference from throughput and packet transfer delay of each terminal station resulting from a power capture effect. CONSTITUTION:Each of terminal stations (1-5) is provided with a packet re- transmission controller (13-53) controlling re-transmission of a packet when the sent packet is not received by an opposite terminal station. The packet re-transmission controller generates uniform random numbers from '1' to 'K' in response to a maximum re-transmission interval K (K is a positive integral number) allocated in advance in each terminal station when a reception reply from the opposite terminal station is not received within a prescribed time and decides the time when the packet is sent again depending on the number thereby controlling the packet re-transmission. The maximum re-transmission interval K is set so that the throughput and the average transfer delay time of each terminal station are identical in the environment taking power capture effect into account.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a packet communication system, and more particularly to a slotted Aloha type packet communication system in which one communication channel is shared by three or more terminal stations for data communication.

[0002]

2. Description of the Related Art Packet communication and non-packet communication are conceivable as means for performing data communication by a large number of terminal stations sharing one communication channel when paying attention to the data transmission mode.

In packet communication, control information such as a destination, a data attribute, a data length, and a serial number is added to data to be transmitted, and the data is transmitted as a burst of data. Further, each terminal station can send data at an arbitrary time or on a time slot with a synchronized system clock as a minimum unit. in this case,
Access the communication channel only when there is a communication request.

On the other hand, in non-packet communication, there is a finite number of communication channels, and an empty channel is allocated by the control of the system side only when the terminal station requests communication. Therefore,
The terminal station occupies the communication channel from the time when the terminal station communication request is accepted until the terminal station issues a communication disconnection request.

As a wireless system using the packet communication system, there are many cases of application to satellite communication, and the first practical application was a computer network called Aloha System of the University of Hawaii, USA. The access protocol used for packet communication has been well studied and many methods have been proposed in the last 15 years. Basically, Aloha method and CSMA (Carrier Sen)
se Multiple Access) method. Below, the slotted
The Aloha method will be described.

In the slotted Aloha system, each terminal station can send a packet only at the beginning of a pre-delimited time interval. As a result, packet collision occurs only when a plurality of terminal stations send out packets at the same time. Therefore, compared with the pure Aloha system, which can send a packet at completely free time, the packet transmission probability is slightly increased, but the packet collision probability is reduced, and the channel capacity is doubled.

Further, when each terminal transmits a packet with a different signal strength, when a packet collision occurs, the packet with the stronger signal strength is correctly received. In this case, since only one retransmission packet is generated due to packet collision, an increase in channel capacity can be expected. Specifically, there is a method in which each terminal station randomly selects the signal strength of the transmitted packet, a method of changing the signal strength according to the importance of the packet, and the like.

[0008]

The above-mentioned conventional packet communication system based on the slotted aloha system has the following problems.

FIG. 5 shows a wireless network composed of five terminal stations, and each terminal station is assumed to be able to communicate with an arbitrary terminal station. For simplicity, it is assumed that each terminal station is arranged in a straight line at equal intervals and transmits radio waves with the same transmission power.

Now, it is assumed that the terminal station 21 and the terminal station 24 simultaneously transmit their respective data to the terminal station 22. Terminal 21
If the distance between the terminal stations 24 and 22 is 1, the distance between the terminal stations 24 and 22 is 2. Therefore, the received input electric field at the terminal station 22 is higher from the terminal station 21 than at the terminal station 24. If each terminal transmits a radio wave by a constant amplitude modulation (for example, FSK modulation), the data of the terminal station 21 is correctly received even though both data of the terminal stations 21 and 24 collide. There is a high probability that Such a phenomenon is called a power capture effect, and many studies have been made.

This power capture effect brings about a favorable result in terms of channel throughput, but when it is applied to an actual system, there arises a problem that the throughput of each terminal station is not uniform due to the geographical condition of each terminal station. ..

For example, the packet generation rate (traffic of only new packets, not including traffic of retransmitted packets) in the idle state of each terminal station (state in which the terminal station does not have a packet waiting to be retransmitted) is the same,
Further, assuming that the new packet generation rate between the terminal stations is also the same, the throughput of the terminal station 23 is the maximum, and the throughput of the terminal stations 21 and 25 at both ends is the minimum, which causes an unfairness.

The phenomenon will be further described as follows. When each terminal has a communication request to other terminals at an equal fixed rate, when this communication request is issued,
If the communication terminal device has an unprocessed retransmission waiting packet, this communication request is rejected, and if not, it is immediately sent out on the communication channel using the next time slot. At this time, if another terminal station simultaneously transmits a packet, packet collision occurs at the receiving side terminal station, and the packet of the transmitting side terminal station closest to the receiving side terminal station is correctly received. The receiving end station returns a reception response to the nearest transmitting end station, and the receiving end station recognizes that its packet has been correctly received and waits for the next communication request. Return to the state.

On the other hand, the other transmission side terminal station knows that its packet transmission has failed at the time when a predetermined waiting time determined by the propagation delay and the hardware delay has passed after transmitting its own packet, Start the resend procedure. Although various methods have been proposed for this retransmission procedure, for example, as the simplest method, the terminal station which recognizes the retransmission of the packet,
A retransmission packet is transmitted with a probability of 1 / K in any time slot within the maximum K time slots counted from that time. After that, repeat the retransmission procedure until your packet is correctly received by the other party.

Here, conventionally, the same maximum retransmission interval K (time slot) was assigned to each terminal station.
Generally, when the value of K is gradually increased, the time (packet delay time) required from when a packet is sent to when it is finally received correctly (packet delay time) increases, and theoretically when the value of K is infinite. This is the same as when not resending.

When packet transmission is controlled by such a communication procedure, the following can be said regarding the throughput, traffic and packet delay time of each terminal station shown in FIG. That is, since the terminal station 23 located in the center of the network has the highest probability of successful transmission and can send new packets one after another, the time ratio that the communication channel is effectively used, that is, the throughput is the highest. high. Furthermore, since the average number of packet retransmissions is the smallest, the average delay time required to transmit one packet is also the smallest. On the other hand, terminal stations 21 and 2 located at both ends of the network
On the other hand, in the case of No. 5, the probability of successful packet transmission is the lowest, and the packet must be retransmitted many times before successful transmission, so the throughput is low and the average packet delay time is large. Further, the terminal stations 22 and 24 located in the middle take a value between these.

As described above, in the conventional packet communication system, the throughput and packet transmission delay of each terminal station are not uniform due to the power capture effect, and it is not possible to give equal communication quality to each terminal station. There is.

An object of the present invention is to provide a packet communication system which can give equal communication quality to each terminal station.

[0019]

The packet communication system of the present invention is a packet communication system of the slotted Aloha system in which one communication channel is shared by three or more terminal stations for data communication. The terminal station has means for retransmitting the packet signal when the reception response from the partner terminal station is not obtained within a predetermined time after transmitting the packet signal. To
The maximum retransmission interval K assigned in advance for each terminal station
A uniform random number from "1" to "K" is generated according to (K is a positive integer), and the retransmission delay time for retransmitting the packet signal is determined by this value. Also,
The maximum retransmission interval K is assigned to each terminal so that the throughput and the packet transmission delay of each terminal become equal based on the relative positional relationship between the terminals.

[0020]

The present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention, showing a case of a wireless communication network composed of five terminal stations 1 to 5 having the same configuration and capable of bidirectional communication.

Each of the terminal stations 1 to 5 converts input / output devices 14 to 54 for inputting / outputting user data and user data to be transmitted into burst data, and then adds information necessary for communication control. Packet controller 12 to 52 that generates a packet signal and converts the received packet signal into user data, converts the packet signal into a radio frequency signal and sends the radio frequency signal to the antenna, and transmits the radio frequency signal received by the antenna. Transceiver 1 for demodulating and outputting a packet signal
1 to 51, and packet retransmission control devices 13 to 53 for controlling the retransmission of the packet that has failed to be transmitted when the transmitted packet is not received by the partner terminal station.

Here, the operation will be described by taking as an example the case where the terminal station 1 and the terminal station 3 communicate with each other.

Now, when the terminal station 1 transmits data to the terminal station 3, the terminal station 2 or 4 close to the terminal station 3 does not simultaneously transmit data, and the terminal station 3 itself transmits data. If not, the data of the terminal station 1 is correctly received by the terminal station 3, converted into user data by the packet control device 32 of the terminal station 3 and sent to the input / output device 34. In this case, that is, when the data transmission from the terminal station 1 to the terminal station 3 is successful, the packet control device 32 sends to the terminal station 1 a reception response indicating that the data has been correctly received. On the other hand, in the terminal station 1, after transmitting the packet, the timer in the packet control device 12 is started, and if the reception response from the terminal station 3 is received within the predetermined waiting time, it is determined that the data transmission is successful, and the data transmission is successful again. It returns to the state of waiting for the input of user data from the input / output device 14.

If the reception response cannot be received within the predetermined waiting time (for example, because the terminal station 1 or 2 simultaneously transmits the radio wave, or the terminal station 3 itself transmits the radio wave). When a packet collision occurs, the packet retransmission control device 13 of the terminal station 1 sets “1” according to the maximum retransmission interval K (K is a positive integer) pre-allocated for each terminal station. To "K" are generated, and the value determines the time to retransmit the packet. For example, if the random number is "5", the packet is transmitted again after 5 time slots counted from the time slot at that time, and thereafter, this operation is repeated until the reception response is received from the terminal station 3 within the waiting time.

By the way, the feature of the present invention is that the maximum retransmission interval Ki (i = 1, 2, 3, 3) is assigned in advance for each terminal station.
4, 5) to generate a uniform random number and determine the time to retransmit the packet. Conventionally, the maximum value of this uniform random number is assigned to each terminal station as the same value. Hereinafter, this Ki determination method will be described using the mathematical model proposed by the present invention. Now, pi: probability that terminal station i will retransmit the packet.

Qi: Probability that terminal station i has a retransmission packet.

Σi: Probability that a new packet will be generated when the terminal station i has no retransmission packet.

Ti: Average round-trip propagation delay time from the terminal station i sending out a packet until receiving a reception response from the partner terminal station.

Ki: The maximum time slot interval until a retransmission is made after the packet transmission of the terminal station i fails.

Si: Throughput of the terminal station i.

Gi: Traffic of terminal station i. Then, pi = 1 / {Ti + (Ki + 1) / 2} (1) gi = (1-qi) σi + qipi (2)
Is. By the way, the terminal station i has to select any one of Ki time slots after the packet transmission fails. This selection method, that is, the random variable Ki
The distribution of is assumed to be uniform. Therefore, if Ti is taken into consideration, the terminal station i is Ti + 1, Ti + 2, ..., T
The packet will be retransmitted in any one of the i + Ki slots. An object of the present invention is to determine the value of Ki so that the throughput and the average transmission delay time of each terminal station are all equal. Here, the average transmission delay time Di of the terminal station i is defined as follows. Di = τi [1+ {Ti + (Ki + 1) / 2} gi / si- (Ki + 1) / 2] (3) where τi represents the packet length of the terminal station i.
In the present embodiment, in order to simplify the analysis, τi is constant for all terminal stations, and its value is assumed to be one time slot. Therefore, when τi = 1, the equation (3) becomes: Di = 1 + {Ti + (Ki + 1) / 2} gi / si− (Ki + 1) / 2 (4).

Further, the equations (1) and (2) are converted into the equation (4).
Substituting into, Di = 1 + Ti + (gi / si-1) / pi
(5)

Next, consider a network in which the number of terminal stations is M. Each terminal station is in one of the following three states. Idle state 1: A state in which there is neither a new packet nor a retransmitted packet. Idle state 2: The terminal station has a new packet with probability σi,
The state to send it. Backlog state: There is a packet collision in the past and is waiting for retransmission.

By the way, it is assumed that the transmission delay of each unretransmitted packet (backlog packet) follows a geometric distribution. That is, each unretransmitted packet is retransmitted in the current time slot with probability pi. Assuming that the packets sent from each terminal station are burst-like, pi> σi holds. After generating a new packet, each terminal cannot generate the next packet until the packet is correctly received. Now,
The probability ρi that the terminal station closer to the receiving terminal station j than the transmitting terminal station i does not transmit the packet at the same time as the time slot transmitted by the terminal station i is defined as follows.

[0036]

Here, M is the number of terminal stations included in the system, and Vi is packet transmission from the terminal station i to the terminal station j, and a packet collision occurs, so that the terminal station i to the terminal station j. It is defined as the risk area where there is a risk that packet transmission to will fail. If the probability that the terminal station is in the idle state 1 is U1, the probability that it is in the idle state 2 is U2, and the probability that it is in the backlog state is U3, then U1 + U2 + U3
= 1 and U1, U2 and U3 are uniquely obtained. U1 = {pi (1-σi) ρi} / {piρi + σi (1-ρi)} (7) U2 = piσiρi / {piρi + σi (1-ρi)} (8) U3 = σi (1-ρi) / {Piρi + σi (1-ρi)} (9) However, it is easily understood from the definition of si and qi that si = U2 (10) and qi = U3 (11). Therefore, equations (10) and (1
If ρi is eliminated from equations (8) and (9) using 1), then piσi {(1-qi) σi-si} = 0 (12). Here, pi ≠ 0 and σi ≠ 0 without impairing generality.
Therefore, si = (1−qi) σi ... (13) is obtained. Further, by substituting the equations (13) and (2) into the equation (5), Di = 1 + Ti + qi / si (14) According to this equation, if the same s
If Ki that gives i and qi is found, Di is the retransmission probability pi
It means that it is determined by the average round trip propagation delay time Ti. In a normal packet radio network, the propagation delay time is negligible compared to the packet length, so that it is possible to give equal throughput and transmission delay time to all terminal stations, which is the object of the present invention.

Next, the throughput is calculated when a different Ki is given to each terminal station with respect to a one-dimensional model in which the number of terminal stations is 5 and all terminal stations are arranged on a straight line at equal intervals.

Throughput Sij from terminal i to terminal j
(R) is defined by the equation (15).

[0040]

Here, n is a terminal station other than the terminal station i in the dangerous area Vr. If the sub total throughput of the terminal station i is Si (r),

[0042]

Here, in Si (r), the terminal i is in the region Rr.
It is defined as the sum of throughputs possessed by all terminal stations j belonging to. This region Rr is a region defined for facilitating mathematical handling of the system model, and specifically, from the position of the terminal station i to the outer contour of the area including all the terminal stations accommodated in the system. It is an area divided by bisectors.

According to the same concept, the total throughput Si that the terminal station i has in the entire area R is defined as the sum of the throughput Si (r) that the terminal station i has for each area Rr obtained by dividing the entire area R. Was

[0045]

It is represented by

Since the throughputs of the respective terminal stations are made equal, it is first assumed that all the terminal stations have the same input traffic, backlog probability and throughput. That is, σi = σo = const. qi = qo = const. si = si = const. And

Solving equation (17) for pi,

[0049]

To obtain. Solving Eq. (18) for all i will give a distribution of pi, which should give equal throughput to all terminals.

Equation (18) is a simultaneous equation containing M unknowns, and at the same time is a transcendental equation. The first thing to consider is the condition for an expression to have a physical meaning. Obviously, the input traffic σo must be greater than so. Otherwise, pi will take a negative value. However, in a network in which the number of end stations that does not retransmit packets is infinite, it is known that the throughput of each end station must be equal to the input traffic of each end station. The mathematical model shown in this example has a solution equivalent to the solution for networks with an infinite number of terminal stations obtained by past research, and considers only the case of σo> so.

Now, in order to make a comparison with the conventional case,
The results of numerical calculations for the distribution of traffic, throughput, and transmission delay time under the same conditions are shown below. In addition,
In the calculation, each variable is set to σo = 0.2, so = 0.1,
It was set to M = 5.

First, the distribution of the retransmission probability pi is obtained by solving the equation (18), and the value of Ki is assigned to each terminal station. Further, the value of K in the conventional method is set to be substantially equal to the average value of the values of Ki in order to meet the operating conditions of this embodiment.

As a result of the calculation, the optimum distribution of Ki is K1 = 1.
2, K2 = 21, K3 = 23, K4 = 21, K5 = 12
Became. Since the average value of these Ki is 17.8, K = 18 was used as the value of K for the conventional method.

The distribution of traffic is shown in FIG. The distribution state of the conventional method is convex upward. This indicates that the terminal station located at the center can transmit new packets more frequently than the terminal stations located at both ends. On the other hand, the distribution state of this method is convex downward. This indicates that the terminal stations located at both ends retransmit the backlog packets more frequently than the terminal station located in the center, and maintain the same throughput as the central terminal station.

The throughput distribution is shown in FIG.
The distribution of the transmission delay time is shown in FIG. In the system according to the present invention, all have flat characteristics, and it is possible to equalize the throughput and the transmission delay time of all the terminal stations included in the network.

[0057]

As described above, according to the present invention, when packet transmission fails at each terminal station, according to the maximum retransmission interval K (K is a positive integer) previously assigned to each terminal station, By generating a uniform random number from "1" to "K", determining a time for retransmitting a packet by the value, and retransmitting the packet, and optimizing the maximum retransmission interval K for each terminal station. Since the difference in the throughput and packet transmission delay of each terminal station caused by the power capture effect can be eliminated and uniformized, equal communication quality can be given to each terminal station.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a diagram showing traffic distribution according to the present embodiment and a conventional method.

FIG. 3 is a diagram showing throughput distributions of this embodiment and a conventional method.

FIG. 4 is a diagram showing transmission delay time distributions in the present embodiment and the conventional method.

FIG. 5 is a diagram showing a wireless network including five terminal stations.

[Explanation of symbols]

 1 to 5 terminal station 13 to 53 packet retransmission control device

Claims (2)

[Claims] 1. A packet communication system of a slotted Aloha system in which one communication channel is shared by three or more terminal stations to perform data communication, and each terminal station transmits a packet signal and then a predetermined signal is transmitted. It has means for retransmitting the packet signal when a reception response from the other end station is not obtained within the time, and this retransmitting means is pre-assigned to each of the end stations when the predetermined time has elapsed. A uniform random number from "1" to "K" is generated according to the maximum retransmission interval K (K is a positive integer), and the retransmission delay time for retransmitting the packet signal is determined by this value. Packet communication system. 2. The maximum retransmission interval K is assigned to each terminal station so that the throughput and packet transmission delay of each terminal station become equal based on the relative positional relationship between the terminal stations. The packet communication system according to claim 1, which is characterized in that.