• No se han encontrado resultados

ANÁLISIS DE ACTORES

In document PLAN OPERATIVO ANUAL -POA- 2017 (página 31-37)

One very important requirement for networking is that correct information arrives at a remote location. Because most information is sent digitally, this means that the bit error rate must be low enough to allow proper function of the networked activity.

5.3.1 Basic Fidelity Techniques

There are several approaches to assuring the fidelity of information sent over a transmission link. You will note that each technique makes trade-offs among data rate, latency, level of fidelity assurance, and system complexity.

You can use majority encoding, which involves sending the data multiple times as shown in Figure 5.6. Let us assume that each data block is sent three times. At the receiver, the received data blocks are compared. If all three agree, the data is passed to an output register. If two of the three agree, their version of the data is passed to the output. If none agree, the data can either be rejected or some arbitrary decision can be made. The fidelity is improved, but the throughput data rate is reduced by a factor of 3 and the output data is delayed by three times the duration of a data block. Sending more repetitions would increase the fidelity in a hostile environment, but would further reduce the throughput rate and increase the latency.

You can also send the data blocks multiple times, but with multiple parity bits added to each block as shown in Figure 5.7. As discussed next, the parity bits for each data block can be checked and any block containing bit errors can be rejected. The first data block received without errors is passed to the output register. In this case, the data throughput rate is reduced and the latency increased by both the percentage of parity bits per block and the number of block repetitions sent. For example, if each block were sent five times and there were 10% parity bits in each block, the throughput data rate would be reduced by a factor of 5.5 and a latency of 5.5 times a data block duration introduced. However, this approach improves the data fidelity. Figure 5.6 Majority encoding requires multiple transmissions of a block of code and the receiver selects the block that is received the same the most times for output.

Figure 5.7 Repetitive transmission with many parity bits requires that each information code block be sent with enough parity bits that a block with errors can be dependably detected. The receiver rejects any block that does not pass parity check and outputs the first error free information block received.

You can retransmit the received data back to the transmitter, check the returned data bit for bit, and repeat the data block if there were any errors as shown in Figure 5.8. Only correct data blocks are placed into the output register, and the transmitter is authorized to send the next data block. If a data block has errors, that block is resent until an error free block is received. This approach assures that every data block will be correctly transmitted (eventually). However, the complexity of a return transmission link is added. Consider a wideband data link from a remote sensor to a control station. Typically, the command link from the control station to the remote sensor has far less bandwidth than the data link; a command link may not even be required. If this fidelity protection approach is used, there must be a link from the control station to the remote sensor, and it must be as wide as the data link. We will be discussing the impact of link bandwidth on network operation later. If the environment does not have significant interference, there will be very little reduction of the data throughput rate or latency with this approach. When there is significant interference or jamming, lots of bit errors can be expected and thus more data block resends, causing decreased data throughput rate and increased latency as a function of the level of hostility of the environment. Figure 5.8 Retransmission data validation requires that each information code block be retransmitted to the transmitter where it is compared with the originally transmitted data. If it is correct an authorization signal is sent to the receiver to allow the code block into the output register. You can add an error detection and correction (EDC) code to each data block as shown in Figure 5.9. If there are errors in a data block, the EDC corrects those errors. This approach is called forward error correction. It provides error free data transmission up to some maximum correctable bit error rate. No return link is required, and the throughput

rate and latency are not changed by the level of hostility of the environment. The amount of reduction of throughput rate and increase of latency depend on the percentage of each data block dedicated to the EDC code; the higher the percentage of code bits, the greater the number of bit errors that can be corrected.

The final approach is to simply increase the transmitter power so that signals are received at a higher signal-to-noise ratio (SNR) and desired signalto-interference ratio. You can get the same effect by reducing the transmitted bit rate which allows a reduced receiver bandwidth. Either of these measures will reduce the received bit error rate, improving the information fidelity. Increased transmitter power can be a significant increase in system complexity, and the reduced data rate will reduce the data throughput rate.

5.3.2 Parity Bits

As discussed above, extra bits are added to the transmitted digital data to protect the information fidelity. This is particularly important in hostile environments with interference including jamming. These extra bits can either be parity bits or an error detection and correction code. Parity bits check that the proper information has been received. The more parity bits provided, the higher the confidence that, if all of the parity bits are received correctly, there were no errors in the received data block. Figure 5.9 Forward error correction requires that an error detection and correction code be placed on each code block. The EDC code is decoded to correct bit errors and the corrected code is output.

5.3.3 EDC

However, an EDC code provides forward error correction. Such a code will detect bad bits (or bytes) and correct them in the received data stream up to some bit (or byte) error rate. The power of the code increases with the number of extra bits or bytes that are added to the data block.

There are two classes of EDC codes. A convolution code is most efficient for randomly spread bit errors. It corrects individual bits. The power of a convolutional code is stated as (n/k), which indicates that there are a total of n output code bits for k information bits. That is, n − k additional EDC code bits are added for each k information bits.

The second class of EDC codes comprises block codes. Block codes correct whole data bytes and are generally more efficient when bit errors come in groups. An example of such a case is a frequency-hopping signal (which you will recall must be digital). If the transmitter hops to a frequency at which there is a strong interfering signal, all of the bits sent at that frequency will be wrong. Actually, there will most likely be close to 50% bit errors. Thus, several contiguous bytes will have many errors. Partial-band jamming is a technique in which some (but not all) of the hopping slots is a technique often used to jam frequency-hopping communication systems. If encountered, it will also cause groups of erroneous bytes when the hopper hops to one of the jammed channels.

The power of a block code is stated as (n, k) meaning that there are n bytes (or symbols) sent for each k information symbol. Thus, n − k extra bytes are added for each n information symbols sent.

An example of a block code is the (31,15) Reed-Solomon code that is used in Link 16, which provides real-time interconnection among airborne, shipboard, and ground military assets. (Note that this code is also used in space broadcasting of compressed television signals.) This specific code can correct (n − k)/2 bad symbols in each n symbol sent. It can also correct one fewer and give an indication to about 10−3 accuracy whether there are any additional uncorrected errors. Because this code sends 31 total bytes for each 15 information bytes, the digital bit transmission rate is more than twice the rate that information bits are sent. In general, this means that over twice the bandwidth is required to send information at any given rate. The advantage is that all of the received bytes will be corrected as long as not more than 8 of 31 bytes contain errors.

5.3.4 Interleaving

When using a block code to protect a frequency hopping link, it is common to transmit a whole block of bytes (i.e., 31 bytes for a 31,15 code) during a single hop. Remember that an occupied hop (i.e., at the frequency of an interfering signal) will cause all of the received bits to be bad. To overcome this problem, the transmitted bytes are interleaved so that not more than 8 of 31 bytes (in this case) will be transmitted at one frequency. Figure 5.10 shows a linear interleaving scheme in which the second 8 bytes are delayed into the next hop, the next into the following hop, and so forth. Thus, no more than 8 contiguous bytes will be lost during an occupied hop. Note that pseudorandom interleaving over a somewhat longer series of bytes is common. Any interleaving approach will cause some increase in latency.

5.3.5 Protecting Content Fidelity

One very important requirement for networking is that correct information arrives at a remote location. Because most information is sent digitally, this means that the bit error rate must be low enough to allow proper functioning of the networked activity.

In document PLAN OPERATIVO ANUAL -POA- 2017 (página 31-37)

Documento similar