Low-complexity decision fusion method for multi-route multi-relay wireless sensor networks

Abstract

A low-complexity decision fusion method for multi-route, multi-relay wireless sensor networks. Each local sensor receives data from the same source via a wireless channel, performs a hard decision on the received data based on the maximum likelihood criterion, and then forwards the decision result to a fusion center via multiple relay nodes. The fusion center extracts a decision metric value from each received signal, which does not require instantaneous channel state information. The fusion center then performs a fusion process to obtain the low-complexity decision metric value required for the decision. The decision metric value is then compared with a decision threshold to obtain the final decision result. This method has the characteristics of low computational complexity and strong robustness.

Description

Technical Field

[0001] The invention belongs to the technical field of wireless communications, and in particular relates to a low-complexity decision fusion method for a multi-route and multi-relay wireless sensor network. Background Art

[0002] A widely distributed information perception network is a key infrastructure for the digitalization, networking, and intelligence of smart cities. The perception layer, at the bottom layer of the network, primarily provides wireless access to local intelligent terminal devices in smart cities. While the network layer accurately and promptly transmits data, the accuracy of data processing at the application layer and the accuracy of data mining conclusions depend on the quality of the perception layer data. Therefore, ensuring reliable data transmission at the perception layer of the ubiquitous perception network is the most effective path to achieving reliable perception of spatiotemporal information in smart cities at the "source of information," making research on this crucial issue extremely important.

[0003] Data from a single sensor can no longer meet the needs of digital and intelligent development in smart cities. Multiple sensors must provide multi-feature observation information for comprehensive, efficient, accurate, and reasonable fusion judgments, estimates, or decisions. Multi-source information fusion technology is an inevitable choice. However, most traditional source information fusion research has not considered the use of relay nodes for collaborative communication. When transmission distances are long, normal information transmission cannot be carried out, which means that communication distance is significantly restricted and network coverage is insufficient. The few existing studies on multi-relay decision fusion require perfect estimation of the instantaneous channel state information (CSI) of each relay channel on each transmission path, which is highly complex to implement and difficult to apply in engineering applications.

[0004] First, in resource-constrained wireless sensor networks, where relay nodes rely on batteries, complex algorithmic calculations performed at the fusion center increase transmission latency and energy consumption, hindering communication system implementation. Therefore, simplifying the log-likelihood ratio (LLR) extraction algorithm at the fusion center to reduce computational complexity becomes crucial. Secondly, in traditional wireless sensor networks, extracting accurate LLR information at the fusion center often requires obtaining channel state information (CSI) along all transmission paths. In practice, due to the dynamic and real-time nature of channel conditions, the CSI estimation process is highly complex and energy-intensive. This contradicts the ideal of low complexity and low cost for wireless sensor networks. Furthermore, errors in the CSI estimation can significantly degrade the performance of the entire coding system. This means that detection systems using accurate CSI are insufficiently robust to CSI. These technical limitations have, to a certain extent, limited the breadth and depth of application of multi-route, multi-relay wireless sensor networks in the transmission of sensor data in emerging smart cities. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides a low-complexity decision fusion method for a multi-route multi-relay wireless sensor network, which uses a binary symmetric channel to achieve low-complexity decision fusion with low computational complexity and strong robustness.

[0006] To achieve the above technical objectives, the adopted technical solution is: a low-complexity decision fusion method for multi-route multi-relay wireless sensor networks, including the following steps:

[0007] Step S1: The data generated by the source is transmitted through the wireless channel and received by the local sensor. The local sensor uses the maximum likelihood criterion to make a hard decision and then forwards the hard decision result to the relay node. The result is then forwarded by multiple relay nodes and sent to the fusion center.

[0008] Step S2: The fusion center extracts the decision metric values that do not require instantaneous channel state information based on the received values obtained from each transmission link, and then performs fusion processing to obtain the low-complexity decision metric values required for the decision;

[0009] Step S3: Compare the low-complexity decision metric value extracted in step S2 with the decision threshold to obtain a final decision result.

[0010] Furthermore, the relay node in step S1 adopts an amplification and forwarding strategy with an amplification factor of 1.

[0011] Furthermore, in step S2, the fusion center has two methods for extracting the decision metric value of each transmission link, namely:

[0012] When the error transfer probability of the relay transmission channel BSC is less than 0.1, the value y received at the fusion center i =1 hour

[0013]

[0014] Receive the value y at the fusion center i =0 o'clock

[0015]

[0016] Among them, P di and P fi Characterize the detection probability and false alarm probability of the i-th local sensor;

[0017] When the error transfer probability of the relay transmission channel BSC is ≥ 0.1,

[0018] Λ2=2y i -1

[0019] Among them, Λ1 and Λ2 represent the fusion center based on the i-th channel received data y under different transmission channel conditions.i The extracted decision metric value no longer contains any instantaneous channel state information.

[0020] Furthermore, the methods for fusing the decision metric values obtained for each transmission link in step S2 are:

[0021]

[0022]

[0023] Wherein, LLR1 represents the low-complexity decision metric value after fusion processing when the decision metric value is Λ1, and LLR2 represents the low-complexity decision metric value after fusion processing when the decision metric value is Λ2.

[0024] Furthermore, the fusion decision method in step S3 is:

[0025]

[0026] Wherein, τ represents the decision threshold, and LLR represents the low-complexity decision metric value, including LLR1 and LLR2.

[0027] Furthermore, the channels between the local sensor and the relay node, between the relay node and the relay node, and between the relay node and the fusion center are all binary symmetric channels.

[0028] The beneficial effects of the present invention are as follows: the present invention provides a low-complexity decision fusion method for a multi-route multi-relay wireless sensor network under a binary symmetric channel, which has the characteristics of low computational complexity, strong robustness and high reliability.

[0029] Compared with the optimal decision metric value, the performance loss of the scheme proposed in the present invention is not large, and no instantaneous channel state information is required. Therefore, it has the characteristics of low implementation complexity, low cost, and strong robustness to channel state information, and can effectively reduce the energy consumption of network nodes. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a data processing flow chart of the fusion center when the error transfer probability of the relay transmission channel is low according to the present invention;

[0031] Figure 2 This is a data processing flow chart of the fusion center when the error transfer probability of the relay transmission channel is high according to the present invention;

[0032] Figure 3 is a flow chart of the data fusion decision-making method of the present invention;

[0033] Figure 4 is a flowchart of the communication system in an embodiment of the present invention;

[0034] Figure 5 BER impact diagram of the embodiment showing a performance comparison between the optimal decision metric value extraction method (18) and the simplified decision metric value extraction method (5) when the relay transmission channel error transfer probability is low;

[0035] Figure 6 is an FER impact diagram comparing the performance of the optimal decision metric value extraction method (18) and the simplified decision metric value extraction method (5) under the condition that the relay transmission channel error transfer probability is low in the embodiment;

[0036] Figure 7 BER diagram of the effect of the number of relays on performance based on the simplified decision metric value extraction method (5) when the relay transmission channel error transfer probability is low in the embodiment;

[0037] Figure 8 This is a diagram showing the influence of the number of relays on the performance of the FER based on the simplified decision metric value extraction method (5) when the relay transmission channel error transfer probability is low in the embodiment;

[0038] Figure 9 BER diagram of the effect of the number of routes on performance based on the simplified decision metric value extraction method (5) when the relay transmission channel error transfer probability is low in the embodiment;

[0039] Figure 10 This is a diagram showing the influence of the number of routes on the performance of the FER based on the simplified decision metric value extraction method (5) when the relay transmission channel error transfer probability is low in the embodiment;

[0040] Figure 11 1 is a diagram showing the effect of the number of relays on the performance of the BER based on the accurate decision metric value extraction method (18) when the relay transmission channel has a high probability of error transfer in the embodiment;

[0041] Figure 12 1 is a diagram showing the BER effect of the number of routes on the performance based on the accurate decision metric value extraction method (18) when the relay transmission channel has a high probability of error transfer in the embodiment;

[0042] Figure 13 BER diagram of the effect of the number of relays on performance based on the simplified decision metric value extraction method (6) when the relay transmission channel has a high probability of error transfer in the embodiment;

[0043] Figure 14 BER diagram of the effect of the number of routes on performance based on the simplified decision metric value extraction method (6) when the relay transmission channel has a high probability of error transfer in the embodiment;

[0044] Figure 15It is a BER impact diagram based on the performance comparison of the simplified decision metric value extraction method (18) and the simplified decision metric value extraction method (6) when the relay transmission channel error transfer probability is high in the embodiment. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Specific implementation method:

[0047] A low-complexity decision fusion method for multi-route multi-relay wireless sensor networks is implemented based on local sensors, relay nodes and fusion centers, such as Figure 3 As shown, the following steps are included:

[0048] A low-complexity decision fusion method for a multi-route multi-relay wireless sensor network includes the following steps:

[0049] S1: Binary data H0 = 0 or H1 = 1 generated by the signal source is transmitted through the wireless channel and received by the local sensor. The local sensor uses the maximum likelihood criterion to make a hard decision and then forwards this hard decision result to the relay node. Multiple relay nodes then forward it to the fusion center. The specific steps are as follows:

[0050] S11: The i-th local sensor receives the binary data H0 or H1 sent by the source, and makes a hard decision based on the maximum likelihood criterion, and the result is x i , x i ∈{0,1};

[0051] S12: Data x i After being forwarded by multiple relay nodes, it is delivered to the fusion center, and the sequence received by the fusion center is y i ;

[0052] x i and y i The relationship is:

[0053]

[0054] Among them, e i,j represents the channel transmission error corresponding to the jth BSC of the i-th link, I represents the total number of transmission links, and J represents the total number of BSCs (binary symmetric channels) from the local sensor to the fusion center. Represents modulo 2 addition.

[0055] In step S12, the relay node adopts an amplification and forwarding strategy with an amplification factor of 1. That is, the local sensor sends data x i , x i After being transmitted by the first BSC, it reaches the first relay node. The data received by this relay node is Since the amplification factor is 1, the relay node does not do any processing and directly Just send it again. After the second BSC transmits, it reaches the second relay node, and the data received by this relay node is The subsequent processing is similar. Finally, when the data reaches the fusion center, the result of formula (1) is obtained.

[0056] S2: The fusion center extracts the decision metric values that do not require instantaneous channel state information (CSI) based on the received values obtained from each transmission link (the entire transmission path from each local sensor to the fusion center), and then performs fusion processing to obtain the low-complexity decision metric values required for the decision.

[0057] In step S2, the fusion center has two methods for extracting the decision metric value of each transmission link. They are:

[0058] When the error transfer probability of the relay transmission channel BSC is small, less than 0.1, that is, when the relay channel quality is good, the value y received at the fusion center i =1 hour

[0059]

[0060] Among them, P di and P fi Characterizes the detection probability and false alarm probability of the i-th local sensor, that is, P di =P(x i =1|H1) and P fi =P(x i =1|H0), which is used to characterize the performance of the signal source to the local sensor.

[0061] Receive the value y at the fusion center i =0 o'clock

[0062]

[0063] When the error transfer probability of the relay transmission channel BSC is large, greater than or equal to 0.1, that is, when the relay channel quality is poor, there is

[0064] Λ2=2y i -1 (4)

[0065] Where Λ1 and Λ2 represent the data y received by the fusion center based on the i-th transmission link under different transmission channel conditions. i The extracted decision metric value no longer contains any instantaneous CSI.

[0066] Furthermore, the fusion processing of the decision metric values obtained in each path in step S2 is respectively:

[0067]

[0068]

[0069] Wherein, LLR1 represents the low-complexity decision metric value after fusion processing when the decision metric value is Λ1, and LLR2 represents the low-complexity decision metric value after fusion processing when the decision metric value is Λ2. Wherein, LLR1 represents the low-complexity decision metric value after fusion processing when the decision metric value is Λ1, and LLR2 represents the low-complexity decision metric value after fusion processing when the decision metric value is Λ2.

[0070] S3: Compare the low-complexity decision metric value extracted in step S2 with the decision threshold to obtain the final decision result.

[0071] Furthermore, the fusion decision method in step S3 is:

[0072]

[0073] Wherein, τ represents the decision threshold, and LLR represents the low-complexity decision metric value, including LLR1 and LLR2.

[0074] The channels between the local sensor and the relay node, between the relay node and the relay node, and between the relay node and the fusion center are all BSCs.

[0075] The theoretical derivation process of the two extraction and processing methods of the decision metric value of each transmission link by the fusion center in step S2 includes the following steps:

[0076] A1: For the transmission link from the local sensor to the fusion center, J serial BSCs can be processed equivalently to obtain an equivalent BSC with an error transfer probability ε i It can be written as:

[0077]

[0078] Among them, ε i,j =P r (e i,j=1) is the error transfer probability of the jth BSC in the i-th link, so the error transfer probability of the equivalent BSC ε i It can be expressed as:

[0079]

[0080] A2: The error transition probability ε of the equivalent BSC is obtained i After explicit expression, based on the maximum likelihood criterion, the initial expression of the optimal decision metric at the fusion center can be written as:

[0081]

[0082] Where y=[y1,y2,…,y I ] is the fusion center receiving vector, which contains the observation values of all I local sensors after J jumps. i ) indicates that the fusion center receives data y based on the i-th channel i The extracted decision metric value of the i-th transmission link.

[0083] In order to obtain the final expression of formula (10), it is necessary to obtain Λ(y i ). The following gives the expression of Λ(y i ) derivation process.

[0084] A3: According to the maximum likelihood criterion, Λ(y i ) can be expressed as:

[0085]

[0086] The proof is:

[0087]

[0088] because

[0089]

[0090] Similarly, we can get

[0091]

[0092] Substituting equations (13) and (14) into equation (12), we can obtain

[0093]

[0094] So far, we have obtained the formula (10) i ), that is, the fusion center receives data y based on the i-th channel i The extracted decision metric expression.

[0095] It is concluded that Λ(y i ), the expression of the optimal decision metric value of the fusion center is given next.

[0096] A5: Substituting equations (16) and (17) into equation (10), the final expression for the optimal decision metric at the fusion center is:

[0097]

[0098] So far, the optimal decision metric expression of the fusion center has been obtained.

[0099] A6: In the process of realizing the optimal decision metric, that is, Equation (18), the fusion center needs to perfectly obtain the instantaneous CSI of each relay BSC in each link, that is, the error transfer probability ε i,j , the implementation complexity is extremely high and difficult to apply in engineering. Even if accurate instantaneous CSI can be obtained, transmitting the obtained CSI from the relay node to the fusion center consumes a lot of energy and other resources. Therefore, it is necessary to simplify and approximate Equation (18) to derive a fusion detection method that does not require instantaneous CSI.

[0100] First, based on formula (18), the theoretical derivation process of the first fusion metric, namely formula (5), is given.

[0101] A7: When the error transfer probability of the transmission channel is low, ε i →0, then formula (18) can be changed to:

[0102]

[0103] At this point, the fusion decision metric value shown in formula (5) is obtained.

[0104] Secondly, based on formula (18), the theoretical derivation process of the second fusion metric, namely formula (6), is given.

[0105] A8: For the hyperbolic tangent function tanhθ and the inverse hyperbolic tangent function arctanhθ, there is the following relationship:

[0106]

[0107] According to formula (20), we can get the equivalent BSC error transfer probability ε given by formula (9): i The following conclusions are drawn:

[0108]

[0109] in,

[0110] A9: According to the tanh criterion:

[0111]

[0112] in,

[0113] A10: When the error transfer probability of the transmission channel is high, according to formula (22), the following equivalent transformation is first performed on formula (16):

[0114]

[0115] According to the derivation of formula (22), formula (23) can be written as:

[0116]

[0117] When the probability of BSC error transfer is high, the following approximate results are obtained:

[0118]

[0119] According to formula (25), formula (24) can be transformed into:

[0120]

[0121] A11: At low signal-to-noise ratio, δ i,j →0, and the exponential function in equation (26) is approximated by the first-order Taylor expansion: Then formula (26) can be written as:

[0122]

[0123] Since log(1+x)≈x, equation (27) can be transformed into

[0124]

[0125] A12: Similarly, formula (17) can be changed to:

[0126]

[0127] A13: According to equations (28) and (29), the final decision metric of the fusion center is:

[0128]

[0129] A14: When each local sensor is the same, that is, P di and P fi When it does not change with the change of i, formula (30) can be changed to

[0130]

[0131] Where K2=(P d -P f ). When P d -P f When >0, discarding the constant term does not affect the result. In this case, Equation (31) can be transformed into:

[0132]

[0133] A15: Equation (32) is similar to the maximum ratio combining metric value in diversity reception. Inspired by this, another decision metric value can be obtained based on equal gain combining:

[0134]

[0135] At this point, the decision metric value shown in formula (6) is obtained.

[0136] like Figure 5 As shown in the figure, the working process of each route of the system is as follows: the source generates binary sequences H0 and H1, which are received by I local sensors through binary channels. Then each local sensor sends the received data to the fusion center through J independent binary symmetric channels. After receiving the signal, the fusion center calculates the received data according to different transmission channel conditions. Figure 1 、 Figure 2 and Figure 3 Process as shown.

[0137] The comparison of the system bit error rate (BER) performance and frame error rate (FER) performance when using the optimal decision metric extraction method (18) and the simplified decision metric extraction method (5) is shown in the figure. Figure 5 and Figure 6 The simulation conditions are set to J = 5 relays, I = 5 and I = 7 routes, respectively. The horizontal axis represents the change in the detection performance of the local sensor. It can be found that the performance of the two methods is roughly the same, which means that when the transition probability of the relay transmission channel is small, the simplified decision metric extraction method (5) obtained by approximate derivation of the exact LLR extraction method (18) reduces the complexity while only causing a small loss in system performance.

[0138] When using the simplified decision metric (5), the effect of the number of relays on the fusion center decision performance is Figure 7 and Figure 8The horizontal axis represents the change in local sensor detection performance. Assume that the number of routers, I, = 7, and the number of relays, J, is 1, 2, 3, 4, and 5, respectively. Overall, we can see that as local sensor performance improves, the system's BER and FER curves show a downward trend, and a threshold phenomenon exists. When the local sensor's performance exceeds the threshold, the curve decreases significantly, indicating a significant performance improvement. Conversely, when the local sensor's performance is below this threshold, the curve decreases more slowly, indicating a smaller performance improvement. We can see that when the relay channel's error transfer probability is low, increasing the number of relays does not significantly change the BER and FER performance.

[0139] Different from the effect of the change in the number of relays on system performance, Figure 9 and Figure 10 The impact of the number of routes on the decision-making performance of the fusion center is given. When we set the number of relays J = 3 and the number of routes I to 1, 3, 5, 7 and 9 respectively, it can be seen that the increase in the number of routes has a more obvious improvement on the system performance.

[0140] When the transfer probability of the relay transmission channel is large and the optimal decision fusion decision metric (18) is used, the impact of the number of relays on the decision performance of the fusion center is Figure 11 Given in Figure 1, let the number of routes I = 7 and the number of relays J be 1, 2, 3, 4, and 5 respectively. The horizontal axis represents the change in the detection performance of the local sensor. Obviously, the channel conditions of relay transmission will become worse as the number of relays increases. When the number of relays is large, the error transfer probability of the equivalent BSC is close to 0.5, so the performance is poor. We simulate the impact of the number of routes I on the decision performance of the fusion center and Figure 12 As shown in , we can see that as the number of routes increases, the performance will also improve.

[0141] The simplified decision fusion metric (6) was simulated and the simulation results were Figure 13 and Figure 14 was demonstrated and Figure 15 The performance of the optimal decision fusion metric (18) and the simplified decision fusion metric (6) were compared. We found that the performance difference between the two methods was minimal, indicating that the performance of our proposed method did not change much after the approximate derivation. Our proposed low-complexity fusion decision scheme does not require CSI estimation, has minimal performance loss, and is less complex. It achieves a better balance between performance and implementation complexity and is particularly suitable for use in multi-route, multi-hop parallel distributed systems.

[0142] In summary, the low-complexity decision fusion method for multi-route multi-relay wireless sensor networks under binary symmetric channels proposed in the present invention has the characteristics of high reliability, strong robustness and low computational complexity.

[0143] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A low-complexity decision fusion method for multi-route multi-relay wireless sensor networks, characterized by: The following steps are involved: Step S1: Data generated by the signal source is transmitted through a wireless channel and received by a local sensor. The local sensor makes a hard decision using the maximum likelihood criterion and then forwards the hard decision result to a relay node. Multiple relay nodes then forward it to the fusion center. The channel between the relay node and the fusion center is a binary symmetric channel (BSC). Each local sensor sends the received data to the fusion center via J independent binary symmetric channels. Step S2: The fusion center first performs equivalent processing on J serial BSCs to obtain an equivalent BSC. Based on the received values obtained from each transmission link, the equivalent transformation between the hyperbolic tangent function and the inverse hyperbolic tangent function, the tanh criterion, and the first-order Taylor expansion are used to extract decision metrics that do not require instantaneous channel state information. The fusion center then performs fusion processing to obtain the low-complexity decision metrics required for decision making. In step S2, the fusion center extracts the decision metric value of each transmission link in two ways: when the error transfer probability of the relay transmission channel BSC is less than 0.1, the fusion center receives the value y i =1 hour Receive the value y at the fusion center i =0 o'clock Among them, P di and P fi Characterize the detection probability and false alarm probability of the i-th local sensor; When the error transfer probability of the relay transmission channel BSC is ≥ 0.1, Λ2=2y i -1 Among them, Λ1 and Λ2 represent the fusion center based on the i-th channel received data y under different transmission channel conditions. i The decision metric value is extracted, which no longer contains any instantaneous channel state information; The methods for fusing the decision metric values obtained for each transmission link in step S2 are: Wherein, LLR1 represents the low-complexity decision metric value after fusion processing when the decision metric value is Λ1, and LLR2 represents the low-complexity decision metric value after fusion processing when the decision metric value is Λ2; Step S3: Compare the low-complexity decision metric value extracted in step S2 with the decision threshold to obtain a final decision result.

2. The low-complexity decision fusion method for a multi-route multi-relay wireless sensor network according to claim 1, characterized in that: The relay node in step S1 adopts an amplification and forwarding strategy with an amplification factor of 1.

3. The low-complexity decision fusion method for a multi-route multi-relay wireless sensor network according to claim 1, characterized in that: The fusion decision method in step S3 is: Wherein, τ represents the decision threshold, and LLR represents the low-complexity decision metric value, including LLR1 and LLR2.

Images