WO2014101341A1

WO2014101341A1 - Mesh ad-hoc network channel adaptive equalizer

Info

Publication number: WO2014101341A1
Application number: PCT/CN2013/071597
Authority: WO
Inventors: 吴传志
Original assignee: 成都泰格微波技术股份有限公司
Priority date: 2012-12-28
Filing date: 2013-02-09
Publication date: 2014-07-03
Also published as: CN103067320A; CN103067320B

Abstract

Disclosed is a MESH ad-hoc network channel adaptive equalizer, comprising: a linear equalizer and a decision feedback equalizer; the linear equalizer comprises a time-delay unit, a tap coefficient unit, an adder, and a sample decider; the tap coefficient unit consists of a noise generation module and a multiplier; the output of each tap coefficient unit is connected to the adder; the equalized output of the adder is connected to the sample decider; the decision feedback equalizer comprises a forward filter and a feedback filter; the feedback filter comprises a tap coefficient unit and a time-delay unit; the equalized output of the adder is connected to the input of the time-delay unit; the output of the time-delay unit is connected to the input of the tap coefficient unit; and the output of each tap coefficient unit is connected to the adder. The present invention has a time-variant characteristic for tracking a communication channel in real time, overcomes the interference between symbols, and effectively avoids signal distortion or error code at a receiving terminal generated during information transmission, thus improving the communication quality of a MESH ad-hoc network.

Description

MESH ad hoc network channel adaptive equalizer

Technical field

The present invention relates to a channel equalizer, and more particularly to an MESH ad hoc network channel adaptive equalizer.

Background technique

With the rapid development of science and technology and the continuous advancement of the information age, communication means are also diversifying. But in fact, there are still some drawbacks in the communication methods that people generally use today. The following are some of the commonly used communication methods and their shortcomings and shortcomings:

(1) Cellular mobile communication network

1 Dependent on infrastructure: mobile terminals need to communicate through fixed base stations, and base stations are connected to backbone switching networks through wired lines, which increases communication costs;

2 The mobile terminal does not have the routing function, and the mobile terminal can only send and receive data through the fixed base station, and the use constraint is large;

3 star-shaped structure, a link fails, a large range of services will be interrupted, and network resistance is poor;

4 high cost of construction, expansion and maintenance;

5 When the mobile terminal is stationary, the data transmission rate can reach 2Mbit/s, but when the mobile terminal moves at high speed, the data transmission rate is only 144kbit/s.

(2) Cluster communication system

1 Similar to a cellular mobile communication network, belonging to a connected network, relying on infrastructure;

2 Generally, it is a private network, mainly based on voice services.

(3) Wireless LAN WLAN

1 The mobile node is equipped with a wireless network card and is connected to the fixed network through the AP access point, depending on the existing network infrastructure of a similar base station or access point;

2 For the network layer, it is a single-hop network and cannot forward data;

3 High-speed communication (802.11b: 11M or 802.11a: 54M) can be achieved within a limited coverage (several hundred meters), but the coverage is limited.

(4) VSAT satellite communication system

1 The coverage is the widest, but the cost is high, the transmission bandwidth is limited, and the transmission delay is large.

(5) moving through

1 Dependent on satellites, when rainy or clouds are heavy, or in special circumstances and where there is occlusion, it is easy to fail, and communication failure occurs;

2 antenna is too heavy, easy to use and carry, must be placed on mobile devices such as cars and ships;

3 There is a long-term homing process that cannot be quickly applied, and the use of many occasions is limited.

In summary, the existing communication networks are mostly based on a stable and reliable communication infrastructure. Once these communication infrastructures are destroyed, conventional communication methods are no longer feasible. It is often at this time that maintaining reliable communication is especially important. In addition to relying on infrastructure, the trunked communication system also has a large limitation on network bandwidth. It uses narrowband technology and bandwidth of 30K. Left and right, data transfer rate 16Kbps Left and right, this makes its data transmission capacity greatly limited. Therefore, in some special occasions, the existing communication means can not meet the communication needs, the biggest problem is that the survivability is too poor, and it is easy to be quickly destroyed.

The MESH ad hoc network communication system has the following features:

1) No center: The status of all nodes in the ad hoc network is equal. It is a peer-to-peer network. Nodes can join and leave the network at any time. When any relay node is disconnected, it will automatically find the nearest communication terminal or repeater. To make up for it, the failure of any node will not affect the normal operation of the entire network, and it has strong invulnerability.

2) Self-organization: The deployment or deployment of the network does not depend on any preset basic communication facilities. After the node is powered on, it can quickly and automatically form an independent network to carry out communication work, which has high communication efficiency and low cost of construction, expansion, maintenance and use. .

3) Multi-hop routing: When a node wants to communicate with a node outside its coverage, it can be implemented by multi-hop forwarding of an intermediate node (communication terminal or repeater).

4) Dynamic topology: The wireless ad hoc communication system allows dynamic changes of its own topology. The topology of the network changes continuously with the changes of the handheld terminal to meet the needs of the call.

5) Smart terminal: The communication terminal is a portable handset or an in-vehicle device, which is convenient to carry and use; in order to save energy, each communication terminal automatically selects the best working mode, and it only keeps in touch with the nearest node to reduce the communication energy consumption.

6) Communication quality: The self-organizing network has strong environmental adaptability, and can communicate with the external network to obtain rich data services; its data transmission has the characteristics of high speed, wide bandwidth, small delay and wide network coverage.

Reliability is a very important indicator in the wireless communication system of the MESH ad hoc network. In a band-limited and time-spread channel, inter-symbol interference due to multipath effects can distort the transmitted signal, which is prone to error in the receiving end, and equalization is a technique for overcoming intersymbol interference. Due to the randomness and degeneration of the wireless channel, the equalizer is required to track the time-varying characteristics of the communication channel in real time. This equalizer is called an adaptive equalizer.

technical problem

The object of the present invention is to overcome the deficiencies of the prior art and provide an MESH ad hoc network that tracks the time-varying characteristics of a communication channel in real time, overcomes inter-symbol interference, avoids signal distortion during information transmission, or generates error at the receiving end. Channel adaptive equalizer.

Technical solution

The object of the present invention is achieved by the following technical solution: an MESH ad hoc network channel adaptive equalizer, which is composed of a linear equalizer and a decision feedback equalizer, and automatically switches between linear equalization and decision feedback equalization through mode selection. And automatically select the best performing algorithm;

The linear equalizer comprises 2N delay units, 2N+1 tap coefficient units, an adder and a sampling decider. The tap coefficient unit is composed of a noise generating module and a multiplier, and the input signals are respectively respectively associated with the first delay. The unit is connected to the input of the first tap coefficient unit, one output of the first delay unit is connected to the input of the second delay unit, and the other output is connected to the second tap coefficient unit, and the output and final stage of the last delay unit The input connection of the tap coefficient unit, the output of each tap coefficient unit is connected to the adder, and the equalized output of the adder is connected with the sampling determiner;

The decision feedback equalizer comprises a forward filter and a feedback filter. The structure of the forward filter is the same as that of the linear equalizer, and the feedback filter comprises at least one tap coefficient unit and a delay unit of the same number. The equalizer output of the adder is connected to the input of the delay unit, the output of the delay unit is connected to the input of the tap coefficient unit, and the output of each tap coefficient unit is connected to the adder.

The linear equalizer of the present invention is a linear LMS equalizer.

Further, the linear LMS equalizer includes a weight setting unit, a delay unit, a tap coefficient unit, an adder, a decision unit, and a calculation error feedback unit, and the input signal is respectively connected to the delay unit, the weight setting unit, the delay unit, and the weight The output of the setting unit is connected to the tap coefficient unit, the output of the tap coefficient unit is connected to the adder, and the output of the adder directly outputs the output signal, and the second output of the adder sequentially passes the decision unit and the training unit and the calculation error feedback unit One input connection, the third output of the adder is connected to another input of the calculation error feedback unit, and the output connection weight setting unit of the error feedback unit is calculated.

The decision feedback equalizer of the present invention is a decision feedback LMS equalizer.

Further, the decision feedback LMS equalizer includes a weight setting unit, a forward delay unit, a forward tap coefficient unit, a feedback delay unit, a feedback tap coefficient unit, an adder, a decision unit, and a calculation error feedback unit, and the input signals are respectively separated. Connected to the forward delay unit and the weight setting unit, the outputs of the forward delay unit and the weight setting unit are connected to the forward tap coefficient unit, the output of the forward tap coefficient unit is connected to the adder, and the output of the adder is directly output. The output signal, the second output of the adder is connected to the training unit through the decision unit, the one output connection of the training unit is connected to calculate the error feedback unit, the other output of the training unit is connected with the input of the feedback delay unit, and the feedback delay unit and feedback The tap coefficient unit is connected, and the output of the feedback tap coefficient unit is connected to the adder; the third output of the adder is connected to the calculation error feedback unit, and the output connection weight setting unit of the error feedback unit is calculated.

Beneficial effect

The beneficial effects of the invention are:

(1) Tracking the time-varying characteristics of the communication channel in real time, overcoming the interference between symbols, effectively avoiding signal distortion during signal transmission or generating bit errors at the receiving end, and improving the communication quality of the MESH ad hoc network;

(2) Adaptive channel equalization based on LMS algorithm, LMS algorithm is a common algorithm, simple structure, moderate amount of calculation, stability is independent of input data and only related to step size, and tracking characteristics of time-varying channel are better than RLS algorithm.

DRAWINGS

1 is a schematic structural view of a linear equalizer;

2 is a schematic structural diagram of a decision feedback equalizer;

3 is a schematic structural diagram of a linear LMS equalizer;

4 is a schematic diagram showing the structure of a decision feedback LMS equalizer.

Embodiments of the invention

The technical solution of the present invention will be described in further detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following.

The MESH ad hoc network channel adaptive equalizer consists of a linear equalizer and a decision feedback equalizer. It automatically switches between linear equalization and decision feedback equalization through mode selection, and automatically selects the best performing algorithm.

As shown in FIG. 1, the linear equalizer includes 2N delay units, 2N+1 tap coefficient units, an adder and a sample decider. The tap coefficient unit is composed of a noise generating module and a multiplier, and the input signals are respectively A delay unit is connected to the input of the first tap coefficient unit, one output of the first delay unit is connected to the input of the second delay unit, and the other output is connected to the second tap coefficient unit, and the output of the last delay unit is Connected to the input of the last-stage tap coefficient unit, the output of each tap coefficient unit is connected to the adder, and the equalized output of the adder is connected to the sampling decider. After the input signal passes through the delay unit, it is multiplied by each corresponding tap coefficient (that is, linearly added), then added to the adder, and finally sent to the sampling decider.

The linear equalizer can be implemented by an FIR filter that takes the sum of the current value of the received signal and the past value as a linear superposition of the filter coefficients as an output.

When the channel distortion is so severe that the linear equalizer is difficult to process, the linear equalizer cannot obtain satisfactory results when there is deep spectrum fading in the channel. At this time, a nonlinear equalizer is needed. The nonlinear equalizer includes a decision feedback equalizer and the maximum Likelihood symbol equalizer and maximum likelihood sequence estimation equalizer.

As shown in FIG. 2, the decision feedback equalizer includes a forward filter and a feedback filter. The structure of the forward filter is the same as that of the linear equalizer. The feedback filter includes at least one tap coefficient unit and the same number of delays. The time unit, the equalizer output of the adder is connected to the input of the delay unit, the output of the delay unit is connected to the input of the tap coefficient unit, and the output of each tap coefficient unit is connected to the adder.

The basic idea of the decision feedback equalizer is that once a signal symbol is detected and determined, the intersymbol interference caused by this symbol can be predicted and eliminated before the subsequent symbols are detected.

The adaptive equalizer is a time-varying filter whose parameters need to be constantly adjusted. The general adaptive algorithm is controlled by error, and the cost function is minimized by the error signal e, that is, the weight of the equalizer is updated in an iterative manner to minimize the cost function. In practical applications, the equalization filter coefficients can be determined by various algorithms. These algorithms mainly include: zero-forcing algorithm, LMS algorithm, RLS algorithm and so on. The criterion of the LMS algorithm is to minimize the mean square error between the desired output value of the equalizer and the actual output value. Find the optimal or near-optimal filter weight by performing the following iterative operation: new weight = original weight + constant * prediction error * current input vector; where prediction error = pre-input expected value - actual output value. In order to better track the real-time changes of the channel, the system periodically transmits a known training sequence, equalizes the channel, and continuously adjusts the filter coefficients to minimize the mean square error.

As shown in Figure 3, the linear equalizer uses a linear LMS equalizer. The linear LMS equalizer includes a weight setting unit, a delay unit, a tap coefficient unit, an adder, a decision unit, and a calculation error feedback unit, and the input signal is respectively connected to the delay unit, the weight setting unit, the delay unit and the weight setting unit. The output is connected to the tap coefficient unit, the output of the tap coefficient unit is connected to the adder, and the output of the adder directly outputs the output signal, and the second output of the adder is sequentially connected to the input unit of the calculation error feedback unit through the decision unit and the training unit. The third output of the adder is connected to another input of the calculation error feedback unit, and the output connection weight setting unit of the error feedback unit is calculated.

As shown in FIG. 4, the decision feedback equalizer is a decision feedback LMS equalizer. The decision feedback LMS equalizer includes a weight setting unit, a forward delay unit, a forward tap coefficient unit, a feedback delay unit, a feedback tap coefficient unit, an adder, a decision unit, and a calculation error feedback unit, and the input signal is forward and forward respectively. The delay unit and the weight setting unit are connected, and the outputs of the forward delay unit and the weight setting unit are connected to the forward tap coefficient unit, the output of the forward tap coefficient unit is connected to the adder, and the output of the adder directly outputs the output signal. The second output of the adder is connected to the training unit through the decision unit, and one output connection of the training unit is connected to calculate an error feedback unit, and the other output of the training unit is connected with the input of the feedback delay unit, and the feedback delay unit and the feedback tap coefficient unit The output of the feedback tap coefficient unit is connected to the adder; the third output of the adder is connected to the calculation error feedback unit, and the output connection weight setting unit of the error feedback unit is calculated.

Since the channel is time-varying, the originator periodically transmits a training sequence to help the equalizer track the channel changes. In MESH, the frame header of each frame of data contains a fixed length pseudo-random sequence (training sequence). The equalizer has two working states: a training mode and a tracking mode. At the beginning of the work, the tap coefficients of the adaptive filter are initialized to a zero vector. The data is processed by the receiver and sent to the adaptive equalizer. The tap coefficients of the equalizer are automatically adjusted under the error control of the received signal and the training sequence. After a certain number of iterations, the equalizer's filter coefficient will be close to the best value that can be achieved and will not change significantly. This state of the equalizer is called convergence. At this time, the equalizer is in the tracking mode (also called the decision mode). In this mode, the tap coefficients of the equalizer are automatically adjusted under the error control of the received signal and the signal constellation. The equalizer is placed after the frame synchronization, and the equalizer automatically switches between the training mode and the tracking mode according to the frame synchronization signal and the training sequence length.

The zero-forcing algorithm amplifies the noise and is difficult to adapt to the deep fading channel. The RLS algorithm has good performance, but the algorithm is complex, the amount of computation is large, and it is not suitable for integration, and its stability depends on the input data. LMS is the most commonly used algorithm. It has a simple structure and moderate computation. The stability is independent of the input data and only related to the step size. The tracking characteristics of the time-varying channel are also better than the RLS algorithm. Therefore, the LMS algorithm is selected in the MESH.

Claims

MESH self-organizing network channel adaptive equalizer, which is characterized in that it is composed of a linear equalizer and a decision feedback equalizer, and automatically switches between linear equalization and decision feedback equalization through mode selection, and automatically selects the best performing algorithm. ;

The linear equalizer comprises 2N delay units, 2N+1 tap coefficient units, an adder and a sampling decider. The tap coefficient unit is composed of a noise generating module and a multiplier, and the input signals are respectively respectively associated with the first delay. The unit is connected to the input of the first tap coefficient unit, one output of the first delay unit is connected to the input of the second delay unit, and the other output is connected to the second tap coefficient unit, and the output and final stage of the last delay unit The input connection of the tap coefficient unit, the output of each tap coefficient unit is connected to the adder, and the equalized output of the adder is connected with the sampling determiner;

The decision feedback equalizer comprises a forward filter and a feedback filter. The structure of the forward filter is the same as that of the linear equalizer, and the feedback filter comprises at least one tap coefficient unit and a delay unit of the same number. The equalizer output of the adder is connected to the input of the delay unit, the output of the delay unit is connected to the input of the tap coefficient unit, and the output of each tap coefficient unit is connected to the adder.
The MESH ad hoc network channel adaptive equalizer according to claim 1, wherein the linear equalizer is a linear LMS equalizer.

[3] The MESH ad hoc network channel adaptive equalizer according to claim 2, wherein said linear LMS equalizer comprises a weight setting unit, a delay unit, a tap coefficient unit, an adder, a decision unit, and The error feedback unit is calculated, and the input signal is respectively connected to the delay unit and the weight setting unit, and the output of the delay unit and the weight setting unit are connected to the tap coefficient unit, and the output of the tap coefficient unit is connected to the adder, and the output of the adder is directly Outputting an output signal, the second output of the adder is sequentially connected to one input of the calculation error feedback unit through the decision unit and the training unit, and the third output of the adder is connected to another input of the calculation error feedback unit, and the error feedback unit is calculated. The output is connected to the weight setting unit.

[4] The MESH ad hoc network channel adaptive equalizer according to claim 1, wherein the decision feedback equalizer is a decision feedback LMS equalizer.

[5] The MESH ad hoc network channel adaptive equalizer according to claim 4, wherein the decision feedback LMS equalizer comprises a weight setting unit, a forward delay unit, a forward tap coefficient unit, and a feedback. Delay unit, feedback tap coefficient unit, adder, decision unit and calculation error feedback unit, the input signal is respectively connected with the forward delay unit and the weight setting unit, and the outputs of the forward delay unit and the weight setting unit are both before Connected to the tap coefficient unit, the output of the forward tap coefficient unit is connected to the adder, and the output of the adder directly outputs the output signal, and the second output of the adder is connected to the training unit through the decision unit, and the output unit of the training unit is calculated incorrectly. The feedback unit, the other output of the training unit is connected with the input of the feedback delay unit, the feedback delay unit is connected with the feedback tap coefficient unit, and the output of the feedback tap coefficient unit is connected to the adder; the third output and calculation error of the adder The feedback unit is connected, and the output connection weight setting unit of the error feedback unit is calculated.