Detection of fast poll responses in a TWACS inbound receiver
View Patent ↗A method for use in a power line communication systems for an electrical distribution system ( 1 ) to quickly and accurately poll electrical meters ( 6 ) installed at user facilities to determine if an outage has occurred at a facility. An outbound communications message is transmitted to the meter at the facility requesting a short response consisting of a bit pattern that is either partially or completely known to the receiver. Any perceived response from the meter is then processed to ascertain whether or not the meter actually transmitted a message. Receipt of a message indicates that an outage has not occurred at that site, while an indication the message was not received indicates an outage has likely occurred. In processing the received message, two types of errors can potentially occur; i.e., a false positive or a false negative. A false positive occurs when a detection algorithm used to process the response indicates that the meter sent a response when actually it did not. A false negative occurs when the detection algorithm indicates the meter did not send a response when actually it did. The method utilizes a signal processing algorithm to determine if a response was actually sent by the meter and is important because it minimizes the probability of both types of errors.
1. In a power line communication system for an electrical distribution network, a fast polling method for quickly and accurately polling electrical meters installed at user facilities to determine whether a meter is connected to the power line comprising:
sending an outbound message over the network to a meter to ascertain whether or not an outage has occurred at the meter's location;
the meter, if it receives the message, sending a fast poll response consisting of a known bit pattern back over the network;
processing the received signal using the known bits in the fast poll response to estimate a plurality of signal quality parameters;
combining values representing the signal quality parameters, using a weighted sum, to produce a single quality parameter; and,
comparing the quality parameter to an adaptively computed threshold to determine whether the response received from the meter was an actual response.
2. The fast polling method of claim 1 in which processing the received signal from the meter includes detecting any unknown bits in the response using a correlation detector algorithm.
3. The fast polling method of claim 2 in which the processing further includes a bit detector which receives samples for each bit in the received signal, certain of the bits having a known bit value, and performs a correlation against data from all the sampled bits, the correlation using samples from the known bits as a reference with respect to bit samples for the unknown bits, an output of the detector being a real number for each bit of the message, a decision as to whether a response to the fast poll message is present or not being made considering all of the detector outputs taken together.
4. The fast polling method of claim 3 in which, in a noisy environment, the correlation detector averages together reference signals for all the bits having a known bit value.
5. The fast polling method of claim 1 in which every bit in the fast poll message is transmitted twice so as to improve bit detection in situations of low signal levels, or high noise levels, thereby to improve reliability.
6. The fast polling method of claim 5 in which processing the response includes summing an output for two transmitted corresponding bits prior to making a decision as to the value of the bits.
7. The fast polling method of claim 2 in which an output vector x of the correlation detector is defined as:
x
=
S
T
S
k
d
k
n
,
where n is the number of reference bits, S is a matrix whose columns contain samples corresponding to bits in the received signal, S T is matrix S transposed, d is a column vector of a given dimension containing the transmitted data, and S k and d k represent a subset of the data whose corresponding bits are already known, and the vector x representing the signal strength for each message bit.
8. The fast polling method of claim 7 in which information is derived from vector x with respect to total signal power of a transmitted fast poll message and the signal-to-noise ratio (SNR) of the message, values for both of these parameters being relatively large if a signal is actually present in the perceived response, but substantially smaller if no signal is present, and the method includes setting a minimum threshold for both the power and SNR of a response signal to eliminate instances where a series of bits accidentally match an expected pattern.
9. In a power line communications system for an electrical distribution network, a fast polling method for quickly and accurately polling electrical meters installed at user facilities to determine if an outage has occurred at a facility comprising:
sending a message over the network to a meter to ascertain whether or not an outage has occurred at the meter's location, the meter, if it receives the message, sending a fast poll response consisting of a known bit pattern;
monitoring the network to detect the response and processing any perceived response to determine if the known bit pattern is present, presence of the known bit pattern indicative of the meter having received the fast poll message and that an outage has not occurred, while absence of the known bit pattern is indicative that the meter did not receive the fast poll message and an outage has occurred, processing of the perceived response including sampling of any bits comprising the response, generating a matrix comprising data obtained from the samples, and correlating the data in the matrix to produce an output vector upon which a determination is made with respect to the outage, the output vector incorporating factors relating to the signal power of the perceived response and its signal-to-noise ratio (SNR), the output vector x produced by the correlation being defined as:
x
=
S
T
S
k
d
k
n
,
where n is the number of reference bits, S is a matrix containing a predetermined number of samples of the transmitted bits, S T is the matrix transposed, d is a column vector of a given dimension containing the transmitted data, and S k and d k represent a known subset of the data represented by the bits, and the vector x representing the signal strength for each message bit; and,
at least one additional metric being used to determine the presence or absence of a signal, the metric being a “max-min” metric used to determine the presence of absence of the signal and defined as:
M
=
max
j
=
1
…
36
{
min
i
=
1
…
8
s
i
,
1
2
,
min
i
=
1
…
8
s
i
,
2
2
,
…
,
min
i
=
1
…
8
s
i
,
36
2
}
where s i,j is an element in row i and column j of matrix S, the metric M being the maximum value in that vector, and the metric M being generally uncorrelated to both signal power and the SNR when a signal is absent, but if a signal is present, the vector being somewhat correlated with signal power.
10. The fast polling method of claim 9 in which the response to the outbound message, if the meter received the message, includes a plurality of bits, some, or all, of which are known to the receiver.
11. The fast polling method of claim 10 in which certain of the bits have a known bit value and processing of the response includes using samples of these bits as a reference for correlating them against bit samples for the perceived response, an output from the correlation being a real number for each bit in the perceived response, a decision as to whether a response signal is actually present or not being made considering all the bit values taken together.
12. The fast polling method of claim 9 in which every bit in the fast poll message is transmitted twice so as to improve bit detection in situations of low signal levels, or high noise levels, thereby to improve reliability.
13. The fast polling method of claim 12 in which processing a perceived response includes summing the received signals corresponding to two repeated bits prior to making a decision as to the value of the bits.
14. In a power line communications system for an electrical distribution network, a fast polling method for quickly and accurately polling electrical meters installed at user facilities to determine if an outage has occurred at a facility comprising:
sending a message over the network to a meter to ascertain whether or not an outage has occurred at the meter's location, the meter, if it receives the message, sending a fast poll response consisting of a known bit pattern;
monitoring the network to detect the response and processing any perceived response to determine if the known bit pattern is present, presence of the known bit pattern indicative of the meter having received the fast poll message and that an outage has not occurred, while absence of the known bit pattern is indicative that the meter did not receive the fast poll message and an outage has occurred, processing of the perceived response including sampling all of the bits comprising the response, generating a matrix S whose columns contain signal samples corresponding to each bit in the response, and computing signal parameters from the data in the matrix to produce an output vector x x upon which a determination is made with respect to the outage, the output vector incorporating factors relating to the quality of the received signal and being defined as:
x x =(( S T S k d k )/ n ) *nSdS k d k x,
where n is the number of reference bits, S is the matrix whose columns contain the signal samples corresponding to each bit in the response and S T is the matrix transposed, d is a column vector of a given dimension containing the transmitted data, and S k and d k represent a known subset of the data represented by the bits, and the vector x represents the signal strength for each message bit; and,
at least one of signal quality metrics being a “max-min” metric used to determine the presence of absence of the signal and defined as:
M
=
max
j
=
1
…
36
{
min
i
=
1
…
8
s
i
,
1
2
,
min
i
=
1
…
8
s
i
,
2
2
,
…
,
min
i
=
1
…
8
s
i
,
36
2
}
where s i,j is an element in row i and column j of matrix S, the metric M being the maximum value in that vector, and the metric M being generally uncorrelated to both signal power and the SNR when a signal is absent, but if a signal is present, the vector being somewhat correlated with signal power.
15. The fast polling method of claim 14 in which certain of the bits have a known bit value and processing of the response includes using samples of these bits as a reference for detecting the any unknown bits, a decision as to whether a response signal is actually present or not being made considering all of the bit values taken together.