CN101835195B - Byzantine failure tolerance method for improving reliability of wireless Mesh backbone network - Google Patents

Abstract

The invention provides a Byzantine fault tolerance method for improving the reliability of the wireless Mesh backbone network. Including the search and establishment of Byzantine units, as well as data backup and fault tolerance. Apply the proposed Byzantine algorithm to improve the existing routing protocol, and select the corresponding node data information for backup in response to different network attack behaviors or abnormal conditions of the node itself. When an attack occurs, some information of the attacked node is tampered with, or a fake node appears in the network, and the node has a human failure, the routing protocol integrated with the Byzantine principle will judge the backup of other nodes in the Byzantine unit where the abnormal node is located. information to obtain correct node information, thereby eliminating interference, enhancing the fault tolerance of the network, and improving the reliability of the wireless Mesh backbone network.

Description

A Byzantine Fault Tolerance Method for Improving the Reliability of Wireless Mesh Backbone Network

technical field

The invention relates to the field of wireless network technology and the field of trusted computing, in particular to a Byzantine fault-tolerant strategy for improving the reliability of a wireless Mesh backbone network.

Background technique

Wireless Mesh Network (WMN for short) is a new type of network structure based on multiple routes and supporting multipoint-to-multipoint technology researched and developed on the basis of mobile Ad hoc network. With the characteristics of management, automatic repair, self-balancing, and mobile broadband, it is a large-capacity, high-speed, and wide-coverage network that can become an effective means of broadband access. Its characteristics of low initial investment, gradual deployment, easy maintenance, robust network and reliable service range will help the wireless mesh network become one of the core communication networks in the future.

The good development prospect of wireless Mesh network makes the research on its reliability particularly important. Apart from hardware factors, a good routing protocol is an important guarantee for its reliability. In terms of the reliability design of wireless mesh network routing protocols, the following factors need to be considered:

(1) Routing criterion. Many existing Ad hoc network routing protocols use the minimum hop count as the routing criterion, but in most cases, the performance of this routing criterion is not optimal in wireless Mesh networks, due to interference conflicts, communication distances, etc. Influenced by other factors, the link quality of the path generated based on this routing criterion will deteriorate, and the end-to-end throughput and bit error rate from the source node to the destination node will become very poor. In order to solve this problem, some new routing criteria are required for wireless mesh networks, and the criteria can correctly reflect the influence of link quality on each index.

(2) Load balancing. In a wireless Mesh network, all nodes share network resources through routing protocols. Therefore, wireless Mesh network routing protocols must meet this requirement of load balancing. For example, when some nodes in the network are congested and become the bottleneck node (Mesh router) of the entire network, new business flows should be able to "bypass" this node and automatically select other paths for data transmission. Routing criteria also need to meet the requirements of load balancing to a certain extent.

(3) Routing fault tolerance: In wireless Mesh networks, routing nodes are relatively fixed, but sometimes routing errors may occur due to data conflicts, and the uncertainty of mobile nodes will also affect the connectivity of the link, and even lead to link failure. This requires the routing protocol of the wireless mesh network to complete routing reconstruction as soon as possible when a routing error occurs, so as to avoid service interruption; or generate a new link due to the emergence of a new available node.

(4) Routing security: If an efficient routing protocol is not well protected, malicious attackers may cause damage to the routing protocol or even completely paralyze the routing protocol. In the wireless Mesh network, because the Mesh client also has the role of routing, Mesh nodes can create, delete or update routing paths in the network according to routing information. This fact is also a common important weakness in wireless Mesh network and Ad hoc network routing protocols, because malicious nodes can generate wrong routing information to change the direction of routing or simply truncate routing. In addition, the attacker can mislead the routing by attacking the routing protocol, thus causing the collapse of the network. Improving the security of routing protocols is also a higher requirement for routing fault-tolerant performance.

In many existing applications, the area where the wireless devices of the wireless Mesh network are arranged is close to the crowd and is vulnerable to human damage, and the network nodes themselves are also extremely vulnerable to attacks, interference, and eavesdropping by other nodes, so security is the key to wireless Mesh networks. One of the important problems that the network must solve. In order to improve the reliability of the wireless Mesh network, it is necessary to improve the existing routing protocol. The current improvement strategy is mainly to improve the security performance of the network protocol through encryption and authentication of legal node information, thereby improving the reliability of the wireless Mesh network. reliability.

The Byzantine principle is a fault-tolerant principle in trusted computing, which enables the system to work normally without being affected even when one or more system components are working abnormally. This principle has a very wide range of applications in trusted computing, fault tolerance, and network security. The essence of the Byzantine principle is to maintain an abstract expression of the data consistency problem in parallel computing. It is realized on the basis of data backup, that is, each component in a system exchanges a certain data or instruction information with each other, and finally The effective part of the system selects the correct data or instructions from the received information on the principle of minority obeying the majority, so as to ensure the reliability of the system.

Contents of the invention

The purpose of the present invention is to provide a Byzantine fault tolerance method for improving the reliability of the wireless Mesh backbone network, which can enhance the fault tolerance capability of the network and improve the reliability of the wireless Mesh backbone network.

The purpose of the present invention is achieved in this way: including the search and establishment of Byzantine units and data backup and fault tolerance;

(1) Search and establishment of Byzantine units

① Take out the NoNnode address and neighbor node address information in the received routing packet;

②The NoNnode address information is added to the routing request packet, that is, when the node broadcasts the routing packet, in addition to sending its own node address information, it also encapsulates the neighbor node address information received by itself into the routing packet and sends it together;

③Traverse the double neighbor node table N-Nform, find out whether the received node information exists in the table, if there is no identical node information, then store the received node information in the double neighbor node table N-Nform;

④ Determine whether the waiting time t is less than T _B , if so, return to step ①;

⑤Traverse the double neighbor node table, and take out three different neighbor node addresses;

⑥Look up the double-adjacent node table, and judge whether the three selected nodes are adjacent nodes to each other. If so, first search the Byzantine unit index table to check whether the node group has been recorded. If not, store the addresses of the three nodes into the index table , and the custom unit number is stored in the table;

⑦ Take the three nodes as the destination nodes, forward the address of itself and the other two nodes except the destination node, and attach the Byzantine unit serial number;

⑧If you receive a node address with a Byzantine unit number, check whether the node group has been recorded in the Byzantine unit index table. If not, store the three node addresses received, together with the Byzantine unit number, into the Byzantine unit index table;

⑨Whether the double-adjacent node list has been traversed, if not, continue to select three different adjacent node addresses, and return to step 5;

⑩The search and establishment of the Byzantine unit is completed;

(2) Data backup and fault tolerance strategy

①Select the own node data information that needs to be backed up;

②Look up the Byzantine unit index table, and send the data information to the three nodes with the same Byzantine unit number;

③ Judging the received node data information, if the source of the message is the same node as the node to which the data information belongs, store the data information in the data information backup table, and record the weight as 1; at the same time, search the Byzantine unit index table to find out the The two nodes with the same Byzantine unit number of the source node forward the data information to the two nodes respectively;

④ If the source of the message is different from the node to which the data information belongs, the information is the node information forwarded by the node to which the information does not belong, and the data information already exists in the data information backup table, then find the information in the data information backup table and assign value plus 1;

⑤ If the received forwarding information is different from the stored data information, then judge whether the weight of the stored information is greater than 1, if yes, then discard the currently received data information; otherwise, store the received data information into the data backup table, the weight is recorded as 1;

⑥Traversing the information backup table, if there are items with the same node name but different data information in the information backup table, then judge the weights of these items, take the data information with high weight as the correct backup information, and delete other information with the same name at the same time; If the weights of these items are the same, use the data information of the latest growth weight as the correct backup information, and delete other information with the same name;

⑦ If the receiving data is not completed, return to step ③;

⑧ end.

According to the structural characteristics of the wireless Mesh network, the present invention firstly proposes the adaptive network structure of the wireless Mesh network applying the Byzantine principle, and then provides a Byzantine algorithm suitable for the wireless Mesh network based on the "oral agreement" solution in the Byzantine principle . The algorithm includes the search and establishment of the Byzantine unit of the wireless Mesh network backbone network node, as well as the storage of selected backup data information and fault-tolerant judgment. Finally, the corresponding strategies for special cases in the practical application of the principle are given.

The definition of technical term involved in the method of the present invention is:

(1) Byzantine Unit (U _B )

If there are 4 nodes in the network, they can reach each other in one hop, that is, they are adjacent nodes to each other, we will form these 4 nodes into a group, which is called a Byzantine unit (see Figure 1).

(2) The Neighbor of the Neighbor Node (NoNnode)

In a wireless Mesh network, if there are three nodes a, b, and c, a is a neighbor of b, b is a neighbor of c, and data information is sent from a to c via b, we define node a as node c Neighbors of neighbors of .

(3) Neighbor and Neighbor From, N-Nform

This table is used to store the neighbor node address and NoNnode address information (Table 1) received by the node.

Table 1 Structure of dual neighbor node table

The data structure of the table is:

struct N-Nform{

char*NoNnode_Address;

char*Neighbor_Node_Address;

};

(4) Byzantine Unit Form (Bform)

This table is used to store the Byzantine unit node address and Byzantine unit number where the node is located (Table 2).

Table 2 Byzantine unit index table structure

The data structure of the table is:

struct Bform{

char*IP;

char*Byzantine_Unit_num;

};

(5) Trigger period (Byzantine Time, TB)

The trigger period refers to the time when the node waits to execute the Byzantine search algorithm since it receives the first routing packet. If the maximum degree of nodes in the wireless Mesh network is N, then the _TB size is set to N times the maximum delay of one-hop data transmission to ensure that the nodes store enough address information during the _TB time.

(6) The weight of data (The quantity of data, QoD)

During the implementation of the Byzantine principle, each node will continuously receive and send data, and the same data may be received multiple times. We define the number of the same data received by the node as the weight of this data.

(7) Data information backup table (Data Form, Dform)

This table is used to store the data information selected for backup, and at the same time record the name of the node to which the data information belongs and the weight of the data (Table 3).

Table 3 Data information backup table structure

The data structure of the table is:

struct Dform{

char*Node_Name;

char*Data;

char*Data_Quantity;

char*Time;

};

The wireless Mesh network adaptive structure among the present invention is:

Applying the principle of Byzantine fault tolerance to network reliability requires certain requirements for the structure of the wireless mesh network. In order to better play the role of the Byzantine principle, considering the mobility of user nodes in the wireless Mesh network, the Byzantine algorithm will only target the backbone nodes of the wireless Mesh network.

Firstly, the adaptive structural constraints of applying the Byzantine principle in the wireless Mesh network are proposed, including the following two points:

①Structure lower limit: the number of network nodes is 4, and all nodes can reach each other with one hop.

② Adaptability requirements: Any node in the network (except the Internet access point) is guaranteed to have at least three neighbor nodes reachable by one hop, and these three nodes can be reached by one hop to each other. That is, any node in the network must be included in at least one "Byzantine unit".

Based on the Byzantine principle, the present invention provides a Byzantine fault-tolerant strategy for improving the reliability of the wireless Mesh backbone network, and uses the proposed Byzantine algorithm to improve the existing routing protocol, and is mainly used to improve the fault tolerance and intrusion tolerance of the wireless Mesh backbone network for nodes. And the ability to tolerate link failure or wrong link status judgment caused by node failure or attack behavior. The improved routing protocol will have the following advantages:

(1) The strategy to improve the reliability of the wireless Mesh network routing protocol is different from the previous node, data encryption and authentication methods, but adopts the method of network nodes performing information backup with each other, without the need for data encryption and decryption process, and the implementation method is simpler fast.

(2) Through the data backup mechanism of the Byzantine algorithm and the improved routing protocol, when a node in the link has data error transmission or packet loss due to link failure, self abnormality or attack, the data transmission can be correct It does not need to choose to retransmit data packets or use new links for transmission like traditional routing protocols. Compared with traditional routing protocols, the improved routing protocol has obvious advantages in execution efficiency and effect.

(3) Through the data fault tolerance mechanism of the Byzantine algorithm, the improved routing protocol can identify the false link failure data packets caused by the attack, thus avoiding unnecessary re-initiation of the link process.

(4) Compared with the improved routing protocol, the fault-tolerant ability is significantly improved (see Figure 6 and Figure 7), so the purpose of improving the reliability of the wireless Mesh backbone network is achieved.

Description of drawings

The basic type of Byzantine unit in Figure 1a and Figure 1b;

Figure 2a-Figure 2c simulation of Byzantine unit establishment process; among them, Figure 2a. WMN backbone node, Figure 2b. One-hop link topology, and Figure 2c. Byzantine unit establishment;

Figure 3 is an example of the process of establishing a Byzantine unit;

Figure 4 The special situation encountered in the practical application of the Byzantine principle;

Figure 5 backup of node information;

Figure 6. Comparison of the packet transmission rate between the improved BAODV protocol and the AODV protocol as the number of network failure nodes increases;

Figure 7. Comparison of the packet transmission rate between the improved BDSR protocol and the DSR protocol as the number of network failure nodes increases;

Figure 8 Byzantine fault tolerance during data transmission;

Figure 9 Wormhole attack;

Figure 10 Wormhole attack after Byzantine fault tolerance;

Figure 11 Byzantine Fault Tolerance for Link Failure Packets (RERR).

Detailed ways

The present invention is described in more detail below in conjunction with accompanying drawing example:

(1) Search and establishment of Byzantine units

The function of the algorithm is to find the Byzantine unit structure in the backbone network nodes of the wireless mesh network, and at the same time, the nodes in the unit record the address information of the unit to prepare for the next data backup (see Figure 2).

The execution of this part of the Byzantine algorithm will be based on the following two types of node information:

①NoNnode address;

②adjacent node address.

It should be noted that the routing data packet of the existing routing protocol does not contain NoNnode address information, so when the algorithm is applied, the routing data packet information of the existing routing protocol must be modified to add the NoNnode address information.

Now the algorithm idea of Byzantine unit search and establishment is described as follows:

① Take out the NoNnode address and neighbor node address information in the received routing packet.

②The NoNnode address information is added to the routing request packet, that is, when the node broadcasts the routing packet, in addition to sending its own node address information, it also encapsulates the neighbor node address information received by itself into the routing packet and sends it together.

③Traverse the double-adjacent node table N-Nform to find out whether the received node information exists in the table (the table elements need to be searched in bit-by-bit correspondence), if there is no identical node information, then store the received node information into the double-neighbor node Form N-Nform.

④ Determine whether the waiting time t is less than T _B , if so, return to step ①.

⑤Traversing the double-adjacent node table, take out three different adjacent node addresses.

⑥Look up the double-adjacent node table, and judge whether the three selected nodes are adjacent nodes to each other. If so, first search the Byzantine unit index table to check whether the node group has been recorded. If not, store the addresses of the three nodes into the index table , and the custom unit number is stored in the table.

⑦ Take the three nodes as the destination nodes, forward the address of itself and the other two nodes except the destination node, and attach the Byzantine unit serial number.

⑧If you receive a node address with a Byzantine unit number, check whether the node group has been recorded in the Byzantine unit index table. If not, store the received three node addresses together with the Byzantine unit number in the Byzantine unit index table.

⑨Whether the double-adjacent node list has been traversed, if not, continue to select three different adjacent node addresses, and return to step ⑤.

⑩The search and establishment of the Byzantine unit is completed;

Now give an example to illustrate the execution process of this part of the algorithm. As shown in Figure 3, the routing request packet is broadcast from point E. After the TB time, with reference to node D, the content of the node's dual-neighbor node table is as follows:

Table 4 The double-adjacent node list of node D

First, take out the three different adjacent node addresses A, B, and C in the table, and then judge whether these three nodes are one-hop adjacent nodes to each other, that is, whether the addresses of any two nodes among the three nodes appear in the table at the same time a row of . Through judgment, it is concluded that the three nodes A, B, and C meet the requirements, so A, B, C and the current node D constitute a Byzantine unit, and D node stores the addresses of A, B, and C nodes in the Byzantine unit index table , generate Byzantine unit numbers at the same time, and forward B, C, D to A; forward A, C, D to B; forward A, B, D node information and Byzantine unit numbers to C.

The pseudo code of the algorithm is as follows:

1) while(t<T _B ) /*Storage and forwarding of dual neighbor node addresses*/

2) receive data, save in RREQ;

3) send Self Address and Neighbor Node Address;

4) if(NoNnode Address||Neighbor Node Address not in N-Nform[])

5) N-Nform[].NoNnode_Address＝NoNnode Address;

6) N-Nform[].Neighbor_Node_Address＝Neighbor Node Address;

7) end if /* Find the address of the current NoNnode and neighbor node in

Whether it exists in N-Nform, if not, it will be stored*/

8) end while

9) while(N-Nform[]!=NULL)

10) select N-Nform[i], N-Nform[j], N-Nform[k];

11) if(N-Nform[i].Neighbor_Node_Address!=

N-Nform[j].Neighbor_Node_Address! =

N-Nform[k].Neighbor_Node_Address)

12) char*A＝N-Nform[i].Neighbor_Node_Address;

13) char*B＝N-Nform[j].Neighbor_Node_Address;

14) char*C＝N-Nform[k].Neighbor_Node_Address;

15) if((AB||BA∈N-Nform[])&&(AC||CA∈N-Nform[])&&

  (BC||CB∈N-Nform[]))/*Determine whether ABC are adjacent nodes*/

16) A, B, C→Bform[].IP;

17) num→Bform[].Byzantine_Unit_num;

18) char*D＝Self Node Address;

19) send B, C, D&num to node A;

20) send A, C, D&num to node B;

21) send A, B, D&num to node C; /*store and forward Byzantine unit information*/

22) end if

23) else

24) break

25) end if

26) end while

27) while(receive IP&num) /*Receive the forwarded Byzantine unit information*/

28) if(IP&num not in Bform[])

29) IP&num→Bform[];

30) end while

(2) Data backup and fault tolerance

After the Byzantine unit is searched and established, it is necessary to carry out the backup process of data information, which is also the main strategy of the Byzantine algorithm to tolerate the abnormality of network nodes.

According to different situations of network attack behavior, select the corresponding node data information for backup. When the attack behavior occurs, some information of the attacked node is tampered with, or when a fake node appears in the network, the routing protocol that incorporates the Byzantine principle will pass Judging the backup information of other nodes in the Byzantine unit where the abnormal node is located, and obtaining the correct node information, thereby eliminating interference.

Now this part of the algorithm is described as follows:

①Select the own node data information that needs to be backed up.

②Look up the Byzantine unit index table, and send the data information to the three nodes with the same Byzantine unit number.

③ Judging the received node data information, if the source of the message is the same node as the node to which the data information belongs, store the data information in the data information backup table, and record the weight as 1. At the same time, look up the Byzantine unit index table, find out two nodes with the same Byzantine unit number as the source node, and forward the data information to these two nodes respectively.

④ If the source of the message is different from the node to which the data information belongs, the information is the node information forwarded by the node to which the information does not belong. This data information already exists in the data information backup table (1-hop information arrives before 2-hop information), Find the information in the information backup table, and add 1 to its weight.

⑤ If the received forwarding information is different from the stored data information, then judge whether the weight of the stored information is greater than 1, if yes, then discard the currently received data information. Otherwise, store the received data information in the data backup table, and record the weight as 1.

⑥ Traversing the information backup table, if there are items with the same node name but different data information in the information backup table, then judge the weights of these items, take the data information with high weight as the correct backup information, and delete other information with the same name. If the weights of these items are the same, the data information with the latest weight increase is used as the correct backup information, and other information with the same name is deleted.

⑦ If receiving data is not completed, return to step ③

⑧ end

The following example illustrates the execution process of this part of the algorithm, as shown in Figure 5, assuming that the current information to be backed up is the IP address of the node. but:

Node 1 searches its own Byzantine unit index table, and sends its own IP address to nodes 2, 3, and M, and node M stores the received IP address of node 1 into the data information backup table, and the weight is recorded as 1, and at the same time Forward the IP address of node 1 to nodes 2 and 3, and node 2 receives the data information sent by node 1, and also receives the IP address of node 1 forwarded by node M, then node 2 stores the data information in the backup table Add 1 to the IP information weight of node 1 (assuming that node 2 has not received the data information forwarded by node 3 at this time). Similarly, nodes M, 2, and 3 also send their own IP addresses to other nodes in the same unit (node M belongs to two Byzantine units, so it will send its own backup information to all 6 nodes of the two units), and receive The nodes to the information also judge, store and forward the information. If there is no abnormality (node failure or external attack) during the entire backup process, taking the Byzantine unit α as an example, the contents of the data information backup table of the unit nodes 1, 2, 3, and M are as follows:

Table 5 Contents of the data information backup table of node 1

Table 6 Contents of the data information backup table of node 2

Table 7 Contents of the data information backup table of node 3

Table 8 Contents of the data information backup table of node M

The pseudo code of the algorithm is as follows:

1) while(data's copy not finish)

2) select A.IP, B.IP, C.IP from Bform[],

where A.Byzantine_Unit_num＝

B.Byzantine_Unit_num＝

C.Byzantine_Unit_num;

3) send self data to node A, node B, node C;

4) while(receive data not finish)

/*Receive backup data from other nodes of the same Byzantine unit*/

5) if(data.name==node.name)

6) save data in Dform[];

7) Dform[].Data_Quantity＝1;

8) select A.IP, B.IP from Bform[],

        where A.Byzantine_Unit_num＝

B.Byzantine_Unit_num＝

node.Byzantine_Unit_num;

9) send data to nodeA, nodeB;

10) end if /*Processing of source node backup data*/

11) else

12) for(i=1; i<=n; i++)

13) if(Dform[i].Node_Name＝＝data.name&&

Dform[i].Data==data)

14) Dform[i].Data_Quantity++;

15) break;

16) end if

17) end for

18) if(i＞n)

19) for(i=1; i<=n; i++)

20) if(Dform[i].Node_Name＝＝data.name&&

Dform[i].Data! =data)

21) if(Dform[i].Data_Quantity＝＝1)

22) save data in Dform[];

23) Dform[].Data_Quantity=1;

24) break;

25) end if

26) end if

27) end for

28) end if

29) if(i＞n)

30) save data in Dform[];

31) Dform[].Data_Quantity=1;

32) end if

33) end else /*processing of forwarding node sending backup data*/

34) select A, B from Dform[],

  where A.Node_Name＝B.Node_Name

and A.time＞B.time; /*data A is newer than data B*/

35) if(A.Data_Quantity<B.Data_Quantity)

36) delete A;

37) else

38) delete B;

39) end while

40) end while

Handling of exceptions

We know that in the case of "oral agreement", if the number of failed individuals is less than 1/3 of the total number of individuals, we believe that such a system can tolerate the Byzantine problem, or that the Byzantine problem can be solved at this time. However, in practical applications, if the Byzantine fault tolerance principle is directly used, there are still special cases when the number of failed nodes is less than 1/3 of the total number of nodes (Figure 4).

We can see from the figure that when the information initiation point is abnormal, the "commander" issued three completely different orders of "attack", "standby" and "retreat" to the three "adjutants". After executing the Byzantine algorithm, the three The instruction set obtained by the "adjutant" is ("attack", "standby", "retreat"), and the Byzantine principle is defined: all loyal "adjutants" obey the same order, then we consider this situation to be a solvable Byzantine The problem, based on this, seems that the three "adjutants" have reached an agreement; but in fact, each of the adjutants is unable to make the correct action.

When a node in the Byzantine unit is abnormal, taking the IP address of the backup node as an example, the abnormal node sends its own IP address to the other three nodes in the same unit. After the nodes exchange information with each other, the other three nodes The correct IP address of the node cannot be obtained, because each node obtains three completely different IP addresses of the node, and the weight of the address information is 1.

In another case, a node in the Byzantine unit does not appear abnormal, but is in a state of motion (although the mobility of the backbone network nodes in the wireless Mesh network is relatively low and relatively fixed, this possibility still exists), so Movement causes some information of the node to change (such as the physical address), so when the node sends changing data information to other nodes in the Byzantine unit during the movement process, the other three nodes cannot get the information that needs to be backed up. The correct information for this node.

Therefore, the proposed algorithm needs to be improved to deal with the situation when the originating point of the backup information is abnormal.

Improvement strategy:

① If a node in the network is in a state of frequent movement, add a regular broadcast information mechanism to the node to increase the frequency of information backup of the node.

② If the nodes in the Byzantine unit cannot obtain the same information of a node for a long time (that is, the weight of the node information is always 1), delete the backup information of the node from the backup information table. Directly regard this node as an invalid node, delete this node and its nodes with the same unit from the Byzantine unit index table, and re-initiate the Byzantine unit search process.

The fault tolerance of wireless Mesh network nodes should not only consider node congestion and failure caused by hardware reasons, but also consider human intrusion factors such as active attacks of mobile nodes, network deception, and congested links.

The reliable routing protocol based on the Byzantine principle will mainly focus on these factors that affect the reliability of the network, and add the Byzantine algorithm to the original routing protocol, so as to achieve the purpose of improving the reliability of the entire network. In order to deal with the attack behavior of nodes, the information we choose to backup nodes can include node ID, routing table, node hop information, link congestion judgment information, etc. Therefore, we select corresponding existing routing protocols for improvement according to different node backup information.

The purpose of proposing the Byzantine algorithm is to generalize the principle of Byzantine fault tolerance, that is, to use the algorithm as a general method to enhance the reliability of existing network routing protocols, and as a supplementary protocol unit. Therefore, for existing practical network protocols (Ad hoc network classic protocol and wireless Mesh network original protocol), we can use it to improve the Byzantine algorithm, so as to improve the reliability of existing routing protocols.

Taking the AODV protocol and the DSR protocol as examples, the Byzantine algorithm is integrated into the two routing protocols, and an improved strategy and an example effect analysis of selecting different backup information are given.

1. BAODV (Byzantine Ad hoc On-demand Distance VectorRouting), an improved protocol for node fault tolerance

(1) Improvement strategy

The Byzantine algorithm first needs to search and establish the Byzantine units of the backbone network nodes of the wireless mesh network. The node information based on this part of the algorithm is:

①NoNnode address;

②adjacent node address.

The process of finding and establishing Byzantine units should be carried out simultaneously with the formation of network topology. Therefore, the information in the route request packet broadcast by the source node of the AODV protocol for finding the route to the destination node is firstly improved.

The routing request packet (RREQ) information includes: source node address, source node serial number, routing request packet serial number, destination node address, destination node serial number and hop value. Now modify the structure of the RREQ packet forwarded by the intermediate node: that is, after the source node address, add the "adjacent node address" and "current node address" that the current node receives the RREQ request.

Then, modify the order in which the AODV protocol judges data packets, that is, when receiving RREQ, first judge whether the node has a route to the destination node or the destination node itself, and if so, judge the number of times the packet has been received , if it is received for the first time, execute the first part of the Byzantine algorithm. During this process, the second and third items of RREQ are taken out, the second item is its NoNnode address, and the third item is its neighbor node address. Store the retrieved information into Bform; then the second and third values of RREQ are modified by the current node, and the node broadcasts the modified RREQ. After the first step of the Byzantine algorithm is completed, continue to execute the original AODV protocol. Algorithmic process. If the packet is received for the second time by the destination node or an intermediate node with a route to the destination node, the AODV protocol is directly executed. If the node is not the destination node, first execute the first part of the Byzantine algorithm, and then execute the AODV protocol.

After the first step of the Byzantine algorithm is executed, the second step of the algorithm is performed: storage of selected backup data information and fault-tolerant judgment. The backup information can be selected according to the actual needs. Of course, it is also possible to back up multiple information of the node at the same time, but it is necessary to modify the data information backup table Dform of the node and add corresponding entries.

In the original AODV protocol, when the node receives different information from the previous backup, it will directly replace the original information with the received information. If the information received by the current node is an error message sent by the message sending node after being attacked If the interference information sent directly by the attack node, the correct information will be replaced by the wrong information. After adding the Byzantine algorithm, when the node wants to change the corresponding record information, it first searches in its own Dform, and then judges the weight of the original backup information. When the weight is greater than 1, the information will not be modified, but the received The information of is stored in Dform, and the weight is recorded as 1. Only when the weight of the newly received information in its Dform is greater than 1 (the second part of the Byzantine algorithm is executed, after the nodes in the Byzantine unit exchange information with each other), the original information will be modified based on the information, and the The backup of the original information in Dform is deleted.

(2) Example analysis

① Fault tolerance of data transmission

During normal wireless Mesh network data transmission, if a node in the link has mistransmission or packet loss due to link failure, self abnormality or attack, it will cause the target node to receive wrong data or even fail to receive it. data. For the case of packet loss, the traditional wireless Mesh network routing protocol is to retransmit the data regularly after receiving the confirmation frame, or select a new link for transmission, and for the case where the destination node receives wrong data, only the destination node The required data will be resent only when the data request is re-initiated, and there is no guarantee that the resent data can reach the destination node correctly.

Based on the above situation, we can use the transmitted data as the backup information of the node when using the Byzantine algorithm.

As shown in FIG. 8 , source node A sends data information to node B, and intermediate nodes of the link are 1, 2, 3, and 4. Normally, when node 2 fails to receive the data sent by node 1 due to link reasons, or node 2 sends wrong information to node 3 due to its own failure or attack, node 3 cannot send the correct data The data message is transmitted to Node B.

After the Byzantine algorithm is used to improve the routing protocol, the transmitted data is used as the backup information of the node. We assume that the correct information is T, and the wrong information is F. When the data T is transmitted to node 1, by executing the second part of the Byzantine algorithm, the node 1. Forward the data T to nodes 2, 3, and 5 in the same Byzantine unit. After receiving the data, nodes 2, 3, and 5 store the data in Dform, and then forward the data. Finally, the data stored in the Dform of node 3 The data are T (weight 2), F (weight 1), because both T and F are data sent by source node A, which belong to different values of information with the same name. After comparing their weights, the data T is sent as the correct information , so that although node 2 transmits wrong information, node B can still receive correct information T.

Of course, when the transmitted data is selected as the backup information, in order to obtain the correct information, the nodes in the Byzantine unit and the link at the same time, except the source of the backup information, must wait for the completion of the backup cycle and make a judgment on the received data The transmission can only be continued after a long time, so the data transmission time will be extended and the transmission efficiency will be reduced.

② Fault tolerance for special attacks

The fault-tolerant situation of special attacks takes Wormhole attack as an example.

Wormhole, wormhole. As the name implies, this kind of attack is that two nodes tunnel the data packet to a non-one-hop node, and then let the data packet continue to be transmitted in the legal link, so that the number of hops between a certain part of the routing nodes in the link does not exceed The illusion of two jumps. As shown in Figure 9, in the AODV protocol, when nodes maintain routes by transmitting Hello messages, attacking nodes A and B respectively transmit the Hello messages of Node 2 and Node 4 to each other through tunnels, thus causing Node 2 and Node 4 Mistakenly think that they are adjacent nodes, but in fact they are not adjacent nodes at all, thus causing errors in node storage information (adjacent node table, etc.), when data is transmitted through the link (link nodes are 1, 2, 4) When , data cannot be sent from node 2 to node 4, because they are not adjacent nodes and there is no one-hop link.

If node 4 forms a Byzantine unit with nodes 5, 6, and 7 (Figure 10), when nodes A and B conduct Wormhole attacks, they will inform node 2 that node 4 is its neighbor node through the transmission of Hello packets, and node 2 will first search for its own neighbors. In the node table, if there is no node 4, node 2 will continue to look up its double neighbor node table N-Nform, if node 4 is indeed its neighbor node, then when the Byzantine unit is established, through the broadcast of the RREQ data packet, the N-Nform of node 2 There must be information (5, 4), (6, 4), (7, 4) in Nform, but node 2 cannot find these information in its N-Nform, so node 2 can judge that "node 4 is its Neighbor node" information is wrong, the node will not be stored in its neighbor node table, and the correct link is still 1, 2, 3, 4.

2. Improved protocol BDSR (Byzantine Dynamic Source Routing Protocol) for link fault tolerance

(1) Improvement strategy

Similar to the idea of improving the AODV protocol, the information in the routing request packet broadcast by the source node of the DSR protocol to find the route is firstly improved.

DSR routing request packet (RREQ) information includes: source node address, destination node address, unique sign number and intermediate node list.

Now modify the structure of the RREQ packet forwarded by the intermediate node: after the source node address, add the "neighboring node address" and "current node address" that the current node receives the RREQ request.

Then, modify the order in which the DSR protocol judges and operates the received data packets: when receiving the RREQ, first judge whether the node is the destination node or has cached the route to the destination node, and if so, then judge whether the packet has been received If it is received for the first time, execute the first part of the Byzantine algorithm, and then continue to execute the original algorithm process of the DSR protocol. If the packet is received for the second time by the destination node or an intermediate node that caches a route to the destination node, the DSR protocol is directly executed. If the node is not the destination node or there is no route to the destination node in its cache, first execute the first part of the Byzantine algorithm, and then execute the DSR protocol.

After the first step of the Byzantine algorithm is executed, the second step of the algorithm is performed: storage of selected backup data information and fault-tolerant judgment. The information to be backed up can also be selected according to actual needs. If multiple pieces of information are backed up at the same time, the data information backup table Dform of the node is correspondingly modified and corresponding entries are added.

After adding the Byzantine algorithm, when the node wants to change the corresponding record information, it first looks for its own Dform, and then judges the weight of the original backup information. If the weight is greater than 1, the information will not be modified, but the received information will be Stored in Dform, the weight is recorded as 1. Only when the newly received information has a weight greater than 1 in its Dform, will the information be used to replace the original information, and the backup of the original information in the Dform will be deleted.

(2) Example analysis

① Fault tolerance of the topology itself

We know that for a wireless Mesh network that satisfies the second structural constraint of the applicability of the Byzantine principle, except for the gateway node connected to the Internet, the backbone nodes in the network are all contained in at least one Byzantine unit. The topology of such a wireless Mesh network backbone node The structure itself is stronger than the general wireless mesh network, with more links available, and the fault tolerance of the network itself is better than that of the ordinary wireless mesh network (as shown in b in Figure 2).

②Link failure caused by node attack

In order to achieve the purpose of monopolizing the resources of a certain link, the attack node will send a fake link failure packet to the predecessor node of the link. During the communication process of the DSR protocol, when the intermediate node receives the false RERR packet sent by the attacking node, it mistakenly believes that the failed link marked in the RERR packet is true, and uses this interrupt link in its own routing table. Delete all the links of the chain, and continue to transmit RERR packets to the source node of the current link, causing the final source node to re-initiate the link search process, and the search process will avoid the failed link marked in the RERR packet.

We use the RERR data packet as backup information, as shown in Figure 11, when node B finds that node 4 is abnormal in the link S→A→1→B→4→5, and the subsequent link is interrupted, it first sends a message to the Byzantine unit where it is located The other nodes 1, 2, and 3 of the other nodes send the RERR, and these nodes forward the RERR to each other after receiving the RERR. The predecessor node 1 of node B in the link stores the RERR in its Dform after receiving the RERR. If the RERR is true, After forwarding each other through the Byzantine unit nodes where B is located, the weight of the RERR packet in the Dform of node 1 must be greater than 1. At this time, node 1 will analyze and process the RERR packet, and then continue to link to the predecessor node A forwards. When the attacking node C sends a forged RERR, node 1 that receives the RERR packet will not receive the same RERR forwarded by other nodes, so the weight of the packet is always 1, so node 1 can make the RERR packet If it is false, the RERR packet is discarded.

Claims (3)

1. a Byzantine failure tolerance method that improves reliability of wireless Mesh backbone network comprises searching and foundation and data backup and fault-tolerant of Byzantium unit, it is characterized in that:

(1) Byzantium unit searching and setting up

1. take out and receive NoNnode address and the neighbors address information in the route bag;

2. add the NoNnode address information in the routing request packet, promptly node is when broadcasting route bag, and except sending self node address information, the neighbors address information of simultaneously self being received is encapsulated in the route bag sends together;

3. whether the two neighbors table N-Nform of traversal search the nodal information of receiving and in table, exist, if there is not same node point information, then the nodal information of receiving are deposited among two neighbors table N-Nform;

4. judge that whether stand-by period t is less than T _B, be then to return the 1. step, T _BBe the triggering cycle;

5. the two neighbors tables of traversal take out three neighbors addresses inequality;

6. search two neighbors tables, each other whether three nodes that judgement is got neighbors, if; At first search Byzantium's unit index table; Check said three nodes record whether, if not then deposit three node addresss in concordance list, self-defined unit number is stored in the table simultaneously;

7. be destination node with three nodes respectively, transmit self and except that destination node other two addresss of node, subsidiary simultaneously Byzantium unit number;

8. as receive the node address of subsidiary Byzantium unit number, check in Byzantium's unit index table this groups of nodes record whether, if do not have, three node addresss then will receiving deposit Byzantium's unit index table in together with Byzantium's unit number;

Whether 9. two neighbors tables have traveled through, if do not have, then continue to choose three different neighbors addresses, return for the 5. step;

10. Byzantium unit searches and sets up end;

(2) data backup and fault-tolerant strategy

1. selection needs self node data information of backup;

2. search Byzantium's unit index table, data message is sent to 3 identical nodes of Byzantium's unit number respectively;

3. the node data information of receiving is judged that if node is same node under informed source and the data message, deposit this data message in the data message backup sheet, weights are designated as 1; Search Byzantium's unit index table simultaneously, find out two nodes identical, data message is transmitted to this two nodes respectively with Byzantium's unit number of informed source node;

If 4. informed source is different with the affiliated node of data message; Then this data message is the nodal information that node is transmitted under the non-information; This data message exists in the data message backup sheet, then in the data message backup sheet, finds this data message, and its weights are added 1;

If the forwarding information that 5. receives, judges then that the weights of stored information whether greater than 1, are then to abandon the current data message of receiving with data information stored is different; Otherwise, depositing the data message of receiving in the data backup table, weights are designated as 1;

6. travel through the information back-up table,, then judge these weights, as correct backup information, delete other information of the same name simultaneously with the high data message of weights if there is identical but the item that data message is different of nodename in the information back-up table; If these weights are identical, then with the up-to-date data message that increases weights as correct backup information, delete other information of the same name;

Do not accomplish if 7. receive data, then turned back to for the 3. step;

8. finish.

2. a kind of Byzantine failure tolerance method that improves reliability of wireless Mesh backbone network according to claim 1 is characterized in that: if certain node in the network is in roving state, then in this node, add the fixed time broadcast information mechanism.

3. a kind of Byzantine failure tolerance method that improves reliability of wireless Mesh backbone network according to claim 1; It is characterized in that: if the node in Byzantium unit can't obtain the identical information of certain node for a long time is that the weights of nodal information are always 1; Then from the backup information table, delete the backup information of this node; Directly this node is regarded as invalid node, this node of deletion reaches and its node with the unit from Byzantium's unit index table, initiates Byzantium's unit search procedure again.

Images