BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a diagram showing a structure of a conventional multihop radio network;
FIG. 2 is a diagram showing a structure of a multihop radio network of embodiments;
FIG. 3 is a diagram showing an exemplary structure using convolutional coding and soft output combining by averaging;
FIG. 4 is a diagram showing simulation results (packet loss rate) according to the exemplary structure of FIG. 3;
FIG. 5 is a diagram showing simulation results (gross traffic) according to the exemplary structure of FIG. 3; and
FIG. 6 is a diagram showing an exemplary structure using convolutional coding and soft output combining based on weights for respective paths.
BEST MODE FOR CARRYING OUT THE INVENTION
Structures of embodiments of the present invention are described using the appended drawings.
FIG. 2 is a diagram showing a general structure of embodiments of the present invention. FIG. 2(a) shows a structure of a network, FIG. 2(b) shows a structure of a transmitter, and FIG. 2(c) shows a structure of a receiver. Respective nodes 210 through 260 configuring a multihop radio network have the structures of the transmitter and the receiver shown in FIGS. 2(b) and 2(c).
As shown in FIG. 2(a), with the embodiments of the present invention, data is transmitted through a part or all of paths (in FIG. 2(a), two paths) in the multihop radio network including multiple paths, and the data received via those paths is diversity combined and decoded, improving characteristics.
In the source node 210 of FIG. 2(a), an encoder 212 performs communication path encoding, and a modulator 216 modulates, as with the prior art. Unlike the prior art, however, a packetizing unit 214 designates relay paths and packetizes outputs from the encoder 212 for the respective paths, whereby signals are transmitted to multiple paths (in this case, two paths). Note that outputs from the encoder 212 may differ for respective paths, or may be the same. For example, bit position may be varied for each of the paths.
The relay nodes 220, 230, 240, and 250 regenerate and relay as with the prior art. The receiver (See FIG. 2(c)) of the destination node 260 demodulates the signals from the multiple paths using a demodulator 262. Thereafter, it depacketizes the packet of binary data hard-decided by a depacketizing/buffering unit 263 and then temporarily stores the depacketized pieces of data for respective paths in a buffer. A reliability data calculator 265 then diversity-combines, taking reliability data into account. A combiner/decoder 267 performs an error correction based on that reliability data.
Since the signals transmitted to the destination node via the relay nodes are regenerated and relayed signals, diversity combination of the hard-decided signals (binary data) is necessary. Therefore, inputs to the reliability data calculator 265 are hard-decided values (binary data). The reliability data calculator 265 calculates likelihood for each of the bits obtained from the multiple paths as reliability information based on the hard decided values for the respective bits, and then outputs it. This output may be soft output, allowing further true representation of likelihood.
FIRST EMBODIMENT
FIG. 3 shows an exemplary structure utilizing convolutional coding as communication path coding and an averaging method as a reliability-based combining method.
In a transmitter shown in FIG. 3(a), input signals are subjected to convolutional encoding by a convolutional encoder 213, the same signals from the convolutional encoder 213 are packetized by the packetizing unit 214 for respective paths, modulated by the modulator 216, and then transmitted to multiple paths (number of paths N).
In a receiver (See FIG. 3(b)), packets transmitted through each of the paths (number of paths N) are demodulated and hard-decided by the demodulator 262, and then depacketized by the depacketizing/buffering unit 263 and temporarily stored in a buffer. A reliability data calculator 275 performs diversity combination of the N number of signals hard decided for each of the paths and stored in the buffer, taking the reliability data into account. More specifically, as shown below, combination taking the reliability data into account is performed by dividing sum of the values for each of the hard-decided paths by the number of paths and then averaging the resulting values.
For example, in the case of transmitting a value +1 using three paths, and data demodulated on the receiver side is +1, +1, and −1, combined output based on the reliability data from the reliability data calculator 275 is +⅓. Viterbi decoding is conducted by a Viterbi decoder 276 using this combined output based on the reliability data. This may be called a soft output combination based on the reliability data.
A simulation experiment is conducted so as to determine advantages of the above-given specific structure. Simulation conditions are shown in the following table.
TABLE 1
|
|
Error Correction Encoder
Convolutional Code
|
|
Encode Rate
½
|
Constraint Length
7
|
Modulation Method
BPSK
|
Communication Path Model
Flat Raleigh Fading
|
Number of Hops
2 (once through relay node)
|
Data Length
500 bits
|
|
FIG. 4 shows packet loss rate characteristics. FIG. 4 shows characteristics of the present invention in cases of using two and three paths, characteristics in a case of using only one path as a conventional method, and characteristics in a case of simply selecting a path from multiple paths through which data is accurately transmitted without performing diversity combination or reliability information calculation. Since diversity combination is performed for multiple paths, characteristic improvement due to using multiple paths can be confirmed. This allows much more improvement than when simply selecting a path through which data is accurately transmitted, and benefits due to the diversity combination can be seen.
Next, gross traffic characteristics are examined to study increase in traffic due to transmission of data through multiple paths. FIG. 5 shows gross traffic characteristics normalized by the number of hops. Gross traffic characteristics result from normalizing the total number of packets transmitted over a network based on the number of hops when transmission of the packets is repeated using automatic repeat requests (ARQ) until no error is detected. For example, the case of accurately carrying out transmission in one time using two paths is counted as two.
According to a graph in FIG. 5, in the case of large signal power, characteristics with the conventional method are better because packets can be successfully transmitted without using multiple paths. However, in the case of small signal power, characteristics with the method using multiple paths are better than with the conventional method. In other words, transmission through multiple paths allows reduction in traffic and decrease in signal power.
As such, since characteristics with multiple paths in terms of the amount of traffic as well as the packet loss rate are better, higher effectiveness of the method of transmitting through multiple paths is confirmed.
SECOND EMBODIMENT
FIG. 6 shows an exemplary structure of convolutional coding as channel coding so as to generate bit streams differing from one other for each of paths, applying a weight based on reliability data for each path, and combining the resulting data.
In a transmitter shown in FIG. 6(a), input signals are convolutional encoded by a convolutional encoder 222, and serial-to-parallel converted by a serial/parallel converter 224, resulting in generation of bit streams for respective transmission paths. The output of an encoder 220 is packetized by a packetizing unit 226 for each of the paths. At this time, CRC coding or parity coding for error detection for each packet is performed.
In a receiver shown in FIG. 6(b), packets transmitted through each path are demodulated and hard-decided by a demodulator 262, depacketized by a depacketizing/buffering unit 285 and then temporarily stored in a buffer. When depacketizing, error detection of the packets is also performed for each path. In a reliability data calculator 280, a weighting factor determination unit 281 determines a weighting factor for each path in accordance with the error detection results. For example, when an error is detected, a weighting factor is set to 0.5, and when an error is not detected, a weighting factor is set to 1. These weighting factors are multiplied for each path by a multiplier 282, and serial-to-parallel converted by a serial/parallel converter 283. A reliability data calculator 280 then outputs the resulting data. Viterbi decoding is conducted by a Viterbi decoder 276 using this combined output from the reliability data calculator 280. In this manner, soft output combination may be performed in accordance with reliability data.
OTHER EMBODIMENTS
Soft combination using reliability data is performed by averaging with the structure of FIG. 3; however, a combined output may be provided using the majority from an odd number of pieces of hard output, taking the reliability data into account. This allows lower error rate for a certain bit than that for bits transmitted through a single path. This majority method may be applied to a case of transmitting through an odd number of paths.
Convolutional coding is used as communication path coding with the structures of FIGS. 3 and 6. Alternatively, turbo coding may be used.
Patent Application
20080026693
