CN103313421A - Back off algorithm in medium access control protocol for multi-hop network and wireless sensor network - Google Patents

Abstract

The invention relates to a back-off algorithm in a medium access control protocol of a multi-hop network and a wireless sensor network. The invention uses different back-off windows for different groups to back off, effectively reducing the time-delay sensitivity of non-time-delay-sensitive groups. The impact of packet competition ensures the effective transmission of delay-sensitive packets. At the same time, for the same type of packets, the node calculates the weighted value of the queue length and transmission times on the basic backoff time slot, so that the longer the queue, the more retransmission times The group has a greater probability to obtain a small backoff time slot, which ensures the fairness of the network.

Description

Backoff Algorithms in Media Access Control Protocols for Multi-Hop Networks and Wireless Sensor Networks

technical field

The invention relates to the technical field of wireless communication, in particular to a backoff algorithm in a medium access control protocol of a multi-hop network and a wireless sensor network.

Background technique

With the development of wireless network, sensor and MEMS technology, the application range of wireless sensor network and ad hoc network is getting wider and wider. Most of the nodes in these networks are powered by batteries, and network throughput and energy efficiency determine the performance and life cycle of the network. In these networks, the medium access control protocol (MAC protocol) determines the use of wireless channels and is used to build the underlying infrastructure of the network system. The current MAC protocol is generally divided into two types, one is based on the wireless channel Time Division Multiple Access (TDMA) protocol, which allocates a fixed wireless channel usage period to each node, thereby avoiding communication between nodes. Mutual interference, but the transmission delay of this protocol is relatively fixed. In many applications, it is necessary to respond to changes in the outside world in a timely manner. Nodes need to report immediately when abnormal conditions are detected instead of waiting for the transmission time slot to arrive before transmitting, and this One protocol has high requirements for time synchronization; the other is a random competition method using wireless channels, and nodes randomly use wireless channels when they need to send data. This random access MAC protocol can respond immediately to external changes, but The key point is to minimize the interference between nodes. In the random access MAC protocol, the nodes share the wireless channel, which causes the collision of data packets. This is because most random access MAC protocols use CSMA/CA and binary exponential back-off. Algorithm (Binary Exponential Backoff, BEB) to reduce the collision, such as the typical IEEE802.11DCF protocol. And this kind of collision is completely backed off by the random backoff time slot generated by the distributed algorithm. The possibility of data packets colliding with each other is relatively high, resulting in retransmission of data packets, which indirectly increases the transmission delay.

For wireless sensor networks and ad hoc networks that transmit heterogeneous data, consider the application scenarios with the following characteristics: the nodes are randomly distributed in an ad hoc network, and the collected data is monitored and transmitted to the sink node through the MAC protocol of random access. In emergencies or special situations, the monitored data is delay-sensitive and needs to be transmitted in time; while the monitored data under normal conditions is not delay-sensitive and can tolerate a certain transmission delay.

For the deployment of nodes in this scenario, the disadvantages of the traditional back-off algorithm are more obvious. Since the delay-sensitive group and the non-delay-sensitive group use the same back-off algorithm, the delay-sensitive group has no priority in forwarding. Relying on the randomly generated backoff time for backoff, it is entirely possible for delay-sensitive packets to lose to non-delay-sensitive packets in channel competition, resulting in excessive transmission delay. In addition, network fairness and energy consumption are also affected by random backoff. Unable to control.

Contents of the invention

The technical problem to be solved by the present invention is to provide a back-off algorithm in the medium access control protocol of the multi-hop network and the wireless sensor network, so that the time-delay-sensitive packets can be transmitted as preferentially as possible over the non-time-delay-sensitive packets.

The technical solution adopted by the present invention to solve the technical problem is: provide a backoff algorithm in the medium access control protocol of a multi-hop network and a wireless sensor network, comprising the following steps:

(1) Divide the packets to be transmitted into two types: delay-sensitive packets and non-delay-sensitive packets, and distinguish them with flag bits in the frame header;

(2) If a node sends a delay-sensitive packet, set the normal back-off window of each node to be (0, CW), and for a non-delay-sensitive packet, the back-off window becomes (0, CW/2), and the node back-off time slot Number B=random(0,CW/2)*(1-qr/Qr)*φ*1/k ^k ; Among them, φ is the delay sensitivity coefficient, qr is the queue length of the current delay sensitive packet of the node, and Qr is The buffer length of the delay-sensitive packet, k is the number of transmissions of the current packet, wherein, the node sends the delay-sensitive packet first;

(3) If a node wants to send a non-delay-sensitive packet, set the normal backoff window of each node to be (0, CW). If the packet type recorded in the network communication is a delay-sensitive packet, then change the backoff window to (CW/2, CW), the number of node back-off time slots B=random(CW/2,CW)*(1-qn/(Qr+Qn))*1/k; if the previously recorded packet type in the network For non-delay-sensitive packets, the number of node back-off slots B=random(0, CW)*(1-qn/(Qr+Qn))*1/k; where, qn is the queue of the node's current non-delay-sensitive packets length, Qr is the buffer length of delay-sensitive packets, Qn is the buffer length of non-delay-sensitive packets, and k is the transmission times of the current packet.

In the step (2), if the node successfully competes for the channel, it will send an RTS packet. After receiving the RTS packet, the parent node will respond to the CTS packet. Both the RTS packet and the CTS packet include the duration of the communication and the communication to be sent. When other neighbor nodes receive the RTS packet and CTS packet, they will set their own sleep timers to go to sleep according to the communication time in them, and record the packet type of the current network communication.

In the step (3), if the node competes for the channel successfully, the sent RTS packet contains NAV to indicate the remaining communication time of this communication, NAV=t*n+T, where n is the number of remaining packets to be sent, and t is The time required to send each packet, T is the total inter-frame interval time; each time a node sends a packet, it waits to receive the confirmation frame from the target node, and resends the packet if it does not receive the confirmation frame.

When the energy of the node is lower than the threshold Eo, the node sends a broadcast message that it is about to fail, and restrains itself from receiving or sending non-delay-sensitive packets, and only receives and sends delay-sensitive packets.

In the step (2), if the node fails to compete for the channel, save the remaining number of back-off time slots Bl for this time, and the new basic number of back-off time slots Bm for the next channel competition, take min(Bl, Bm) as the new Number of backoff slots for a round.

In the step (3), if the node fails to compete for the channel, the regenerated random number is used as the basic backoff time slot number for a new round.

During the transmission process, if the node itself collects and generates another packet, it must re-participate in the next channel competition after the end of this communication.

Beneficial effect

Owing to adopting above-mentioned technical scheme, the present invention has following advantage and positive effect compared with prior art:

With the present invention, when the node has no delay-sensitive packets to send and knows that there may be delay-sensitive packets in the network, the backoff window is automatically delayed to (CW/2, CW), completely avoiding collisions with delay-sensitive packets, even if The node has not detected the delay-sensitive packets in the network, and the backoff window (0, CW) is twice the delay-sensitive packet backoff window (0, CW/2), and the latter has a greater probability of obtaining a smaller random number.

With the present invention, when the basic back-off time slots are similar among groups of the same type, the longer the queue length and the more retransmission times, the easier it is to obtain smaller back-off time slots, which ensures the fairness of the network.

Utilizing the present invention, among different groups, the weighting coefficients of delay-sensitive packets are smaller than those of non-time-delay-sensitive packets, unless the non-time-delay-sensitive packets do not detect the transmission of time-delay-sensitive packets in the network The full backoff window produces a basic backoff slot smaller than that of the delay sensitive packet. After weighted calculation, the final backoff slot may also be larger than that of the delay sensitive packet, which is beneficial to the priority transmission of the delay sensitive packet.

Description of drawings

Fig. 1 is the algorithm flowchart of the present invention;

2 to 4 are schematic diagrams of specific examples of the present invention.

Detailed ways

Below in conjunction with specific embodiment, further illustrate the present invention. It should be understood that these examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the teachings of the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of the present application.

Embodiments of the present invention relate to a backoff algorithm in a medium access control protocol of a multi-hop network and a wireless sensor network, as shown in FIG. 1 , comprising the following steps:

(1) The node adopts a periodic sleep and wake-up mechanism. When the node wakes up, it first listens to the channel. If the channel is busy and the node itself has no packets to send, the node continues to listen for a period of time and still has no packets to send or receive. , directly set the timer to enter the sleep state, and wait for the arrival of the next wake-up period; if the node has a packet to send, it will continue to listen until it receives a packet containing the communication time (the node may wake up in the middle of the transmission stage of a certain packet At this time, it can only detect that there is energy in the channel but cannot resolve what packet it received) or detect that the channel is idle, and then set the sleep timer according to the NAV value to go to sleep or start to compete for the channel.

(2) Divide the packets to be transmitted into two types: delay-sensitive packets and non-delay-sensitive packets, and distinguish them with flag bits in the frame header; delay-sensitive packets are always forwarded prior to non-delay-sensitive packets. When the node detects that the channel is idle, it starts to compete for the channel and execute the backoff algorithm. If the node detects the control packets of other nodes during the back-off period, the competition fails, the back-off process ends, the packet type and communication time transmitted by other nodes are recorded, and it goes to sleep. When the communication time of other nodes arrives, wake up and re-compete for the channel.

(3) The process of nodes sending delay-sensitive packets is as follows:

(a) When a node wants to send delay-sensitive packets, it reduces its own back-off window from (0, CW) to (0, CW/2), and randomly selects the number of basic back-off slots B=random(0, CW/ 2) When the node backoff fails, the remaining backoff time slots this time are B1(k), and the number of backoff time slots generated by the node re-competing for the channel next time is Bm(k+1), and the new round of basic backoff time slots is taken B(k+1)=min(B1(k), Bm(k+1)).

(b) Multiply the number of basic backoff slots by the weight coefficient of the number of transmissions to get B(k)=B*1/k ^k , where k represents the number of ongoing transmissions of a packet. It can be seen from the formula that when the number of packet transmission failures increases, the value of k As B increases, B is a monotonically decreasing function of k, and it converges to 0 very quickly. The value of B(k) will obviously tend to become smaller when nodes compete again next time, and there will be a greater chance of successfully transmitting packets.

(c) Multiply the number of backoff slots obtained by the queue length weight coefficient B(k,q)=B(k)*(1-qr/Qr)*φ, qr is the queue of the node's current delay-sensitive packet length, Qr is the buffer length of the delay-sensitive packet. B(k,q) is the final number of back-off time slots. It can be seen from the formula that the longer the queue length is, the smaller the value of B(k,q) is, and the easier it is to transmit, which ensures that nodes with longer queues have greater Network fairness for probabilistically transmitted packets. φ is the delay sensitivity coefficient, which is to ensure that the queue weight of delay-sensitive packets is not greater than the queue weight of non-delay-sensitive packets.

(4) The process of nodes sending non-delay-sensitive packets is as follows:

(a) When a node wants to send a non-delay-sensitive packet, it first checks the recorded packet type in the last network communication. If it is a delay-sensitive packet, it thinks that there may still be delay-sensitive packet competition in this communication, and sends its own The backoff window is changed from (0, CW) to (CW/2, CW) to "give way" to delay-sensitive packets as much as possible. Even if there is no delay-sensitive packet communication in the network this time, it will not affect the competition with other non-delay-sensitive packets, because all competing nodes have changed the window because of the recorded packet type. If the recorded packet type of the last communication is a non-delay-sensitive packet, it is considered that this communication does not need to change the contention window, which is still (0, CW). The number of basic backoff slots B=random(0, CW) or (CW/2, CW).

(b) Multiply the number of basic backoff slots by the weight coefficient of the number of transmissions to get B(k)=B*1/k, where k represents the number of ongoing transmissions of a packet. It can be seen from the formula that when the number of packet transmission failures increases, the value of k increases. B is a monotonous decreasing function of k. Compared with similar groups, the more times of competition failures, the easier it is for the next group to be sent successfully. However, compared with delay-sensitive groups, the convergence speed of its weighting coefficient on k is obviously weaker The weighting coefficient of the latter is 1/(k+1) ^k , and the weighted backoff slot of the delay-sensitive packet tends to 0 faster.

(c) Multiply the number of backoff slots obtained by the queue length weight coefficient B(k, q)=B(k)*(1-qn/(Qr+Qn)), where qn is the current non-delay-sensitive node The queue length of the packet, Qr is the buffer length of the delay sensitive packet, and Qn is the buffer length of the non-delay sensitive packet. B(k,q) is the final number of back-off slots. It can be seen from the formula that compared with the same type of packets, the longer the queue length, the smaller the value of B(k,q), and the easier it is to transmit, ensuring a longer queue The nodes have a greater probability of transmitting packets for network fairness. But compared to delay-sensitive packets,

To ensure that (1-qr/Qr)*φ<1-qn/(Qr+Qn-qn)/(Qr+Qn);

Let φ=Qr/(Qn+Qr) 0<φ<1→(1-qr/Qr)*φ=(Qr-qr)/(Qr+Qn);

Qr-qr<Qr+Qn-qn→qn<Qn+qr;

Because qn≤Qn, qr≥1→qn<Qn+qr is always established.

So (1-qr/Qr)*φ<1-qn/(Qr+Qn) is always established.

It is ensured that the queue weight coefficient of the delay sensitive packet is smaller than the queue weight coefficient of the non-delay sensitive packet.

Compared with non-delay-sensitive packets, if the number of basic back-off slots of the former is less than or equal to the number of basic back-off slots of non-delay-sensitive packets, the final back-off slots must be smaller, even if the former’s basic back-off slots The number of time slots is greater than the number of back-off time slots of the latter, and the final back-off time slot may also be smaller than the back-off time slot of non-delay-sensitive packets.

Assume that nodes 1, 2, 3, and 5 in Figure 2(a) all want to communicate with central node 0 in the first round of communication. After the competition starts, the channel is idle and no packets are sent in the channel, and each node starts the backoff process. Each node generates its own basic back-off time slot. Since nodes 1 and 3 contain delay-sensitive packets, the back-off window is (0, CW/2), and other nodes do not know the existence of this communication in the network for the time being. Backoff is still performed according to the normal window (0, CW). Set CW=64, the basic back-off time slots generated by each node are 11, 32, 21, 20 (values generated completely randomly), according to the back-off algorithm, set Qr=20, Qn=20, the final back-off time slot of each node for:

B1=11*1/2*(1-4/20)*20/(20+20)=2

B2=32*1*(1–1/40)=31

B3=21*1/2*(1–2/20)*20/(20+20)=4

B5=20*1*(1-5/40)=17

Therefore, node 1 will compete to win the channel and send an RTS packet in the second time slot. Node 2 and node 6 receive the RTS packet within the communication range of node 1, end the backoff setting timer and go to sleep; the central node receives the RTS After grouping, send a CTS packet to respond to node 1, and nodes 3, 4, and 5 go to sleep after receiving the CTS packet, as shown in Figure 2(b). Then node 1 sends data packets until the end of the communication, as shown in Figure 2(c). The first round of communication is over.

Assuming that in Figure 3(a) after the start of the second round of communication, each node wakes up and starts to compete for the channel after the communication of node 1 ends, node 1 has no packets to send, and node 4 newly generates 3 non-delay-sensitive packets, The grouping of other nodes remains unchanged. At this time, because nodes 2, 4, 5, and 6 recorded the transmission of delay-sensitive packets last time in the network, the backoff window becomes (CW/2, CW), so this communication must be the successful channel of node 3 competition, according to The previous proof, because the basic random backoff number generated by it is between (0, CW/2), which is smaller than the basic backoff time slot of non-delay-sensitive packets, the time slot is also the smallest after multiplying the weighted value, as shown in Figure 3 (b), (c) shown.

Suppose that in Figure 4(a) after the start of the third round of communication, each node wakes up and starts to compete for the channel after the communication of node 1. Node 2 newly generates 2 non-delay-sensitive packets, and node 6 generates 4 new packets. For non-delay-sensitive packets, the packets of other nodes remain unchanged. Since the last round of communication in the network is still delay-sensitive packets, the backoff window of each node is (CW/2, CW), and the basic backoff numbers generated by each node are 40, 57, 60, 63, 51, respectively. Nodes 2 and 5 are the third competition channel, node 4 is the second competition channel, node 6 is the first competition channel, and node 3 is also the first competition because it is the first time to send non-delay sensitive packets. channel. The final backoff time slot of each node is:

B2=40*1/3*(1-3/20)=11

B3=57*1*(1-2/20)=51

B4=60*1/2*(1-3/20)=25

B5=63*1/3*(1-2/20)=18

B6=51*1*(1-4/20)=40

Therefore, node 2 will compete to win the channel, as shown in Figure 4(b) and (c), from the calculation results, although the basic backoff slot of node 5 is larger, the final backoff slot number is still smaller than that of nodes 3, 4, and 6, because Node 5 has more transmission times than them, so the weight of transmission times is very small, so that the final number of backoff time slots is still smaller than them, reflecting the fairness of the network.

The following rounds of communication will not be explained one by one. If there is no new delay-sensitive packet in the next round, the backoff window of each competing node will be restored to (0, CW), and other processes will remain unchanged.

Claims (7)

1. a backoff algorithm in the medium access control protocol of multi-hop network and wireless sensor network, is characterized in that, comprises the following steps: (1) Divide the packets to be transmitted into two types: delay-sensitive packets and non-delay-sensitive packets, and distinguish them with flag bits in the frame header; (2) If a node sends a delay-sensitive packet, set the normal back-off window of each node to be (0, CW), and for a non-delay-sensitive packet, the back-off window becomes (0, CW/2), and the node back-off time slot Number B=random(0,CW/2)*(1-qr/Qr)*φ*1/k ^k ; Among them, φ is the delay sensitivity coefficient, qr is the queue length of the current delay sensitive packet of the node, and Qr is The buffer length of the delay-sensitive packet, k is the number of transmissions of the current packet, wherein, the node sends the delay-sensitive packet first; (3) If a node wants to send a non-delay-sensitive packet, set the normal backoff window of each node to be (0, CW). If the packet type recorded in the network communication is a delay-sensitive packet, then change the backoff window to (CW/2, CW), the number of node back-off time slots B=random(CW/2,CW)*(1-qn/(Qr+Qn))*1/k; if the previously recorded packet type in the network For non-delay-sensitive packets, the number of node back-off slots B=random(0, CW)*(1-qn/(Qr+Qn))*1/k; where, qn is the queue of the node's current non-delay-sensitive packets length, Qr is the buffer length of delay-sensitive packets, Qn is the buffer length of non-delay-sensitive packets, and k is the transmission times of the current packet. 2. The back-off algorithm in the medium access control protocol of multi-hop network and wireless sensor network according to claim 1, characterized in that, in the step (2), if the node successfully competes for the channel, it sends an RTS packet, and the parent After the node receives the RTS packet, it responds to the CTS packet. Both the RTS packet and the CTS packet include the duration of the communication and the type of packet to be sent in this communication; when other neighbor nodes receive the RTS packet and the CTS packet, they will The communication time sets its own sleep timer to go to sleep, and records the grouping type of the current network for communication. 3. The back-off algorithm in the medium access control protocol of multi-hop network and wireless sensor network according to claim 1, characterized in that, in the step (3), if the node competes for the channel successfully, the RTS packet sent contains NAV It is used to indicate the remaining communication time of this communication, NAV=t*n+T, where n is the number of remaining packets to be sent, t is the time required to send each packet, and T is the total inter-frame interval time; node After sending a packet each time, it waits to receive the confirmation frame of the target node, and resends the packet if the confirmation frame is not received. 4. the back-off algorithm in the media access control protocol of multi-hop network and wireless sensor network according to claim 1, is characterized in that, when the energy of described node is lower than threshold value Eo, node sends a broadcast that self is about to fail information, and inhibit itself from receiving or sending non-delay-sensitive packets, and only receive and send delay-sensitive packets. 5. The backoff algorithm in the medium access control protocol of multi-hop network and wireless sensor network according to claim 1, characterized in that, if the node fails to compete for the channel in the step (2), the remaining backoff of this time is saved The number of time slots Bl is the new basic number of back-off time slots Bm generated in the next channel competition, and min(Bl, Bm) is taken as the number of back-off time slots for the new round. 6. The back-off algorithm in the medium access control protocol of multi-hop network and wireless sensor network according to claim 1, characterized in that, if the node fails to compete for the channel in the step (3), the newly generated random The number is used as the number of backoff slots for the new round. 7. The back-off algorithm in the medium access control protocol of multi-hop network and wireless sensor network according to claim 1, is characterized in that, if node itself collects and produces other grouping in transmission process, then will be in this communication After the end, participate in the next channel competition again.

Images