Energy management for wireless sensor networks

Abstract

This invention concerns remote sensor networks, and particularly energy management for wireless sensor networks. In a first aspect the invention is a wireless sensor node specified to operate for a given lifetime, comprising: an onboard computer system and a set of one or more associated sensors. The computer system operates to periodically sample data from each sensor of the set of associated sensors, and to store a multi-state model representing one or more phenomena described by the collected data. And, the computer system operates to calculate a value associated with movement of the phenomena between the states of the multi-state model, and to adjust the rate of sampling of one or more of the set of associated sensors depending on the calculated value. In other aspects the invention is a network of sensor nodes and a method of operation.

1 . A wireless sensor node configured to operate for a given lifetime, comprising:
 an onboard computer system; and   a set of one or more associated sensors;   wherein, the computer system is configured to:
 periodically sample data from each sensor of the set of associated sensors, 
 store a multi-state model representing one or more phenomena described by the collected data; 
 calculate a value associated with movement of the phenomena between the states of the multi-state model; and 
 adjust the rate of sampling of one or more of the set of associated sensors depending on the calculated value. 
   
     
     
         2 . A wireless sensor node according to  claim 1 , wherein the multi-state model includes an entropy for each state, defining the average information contained in the phenomena when in that state. 
     
     
         3 . A wireless sensor node according to  claim 1 , wherein the multi-state model also includes a probability mass function (PMF) for each state to describe a likelihood of a measurement returning a particular value while the phenomena is in that state. 
     
     
         4 . A wireless sensor node according to  claim 1 , wherein the multi-state model further includes a transition weight for each respective transition between states, defining the likelihood of each transition; or no transition. 
     
     
         5 . A wireless sensor node according to  claim 1 , wherein the computer system is configured, during each cycle of operation, to:
 collect a set of fresh data from the set of sensors;   using the fresh set of data to generate a new likelihood value for the most likely state the phenomena is in;   calculate a value representing an index of surprise associated with movements of the phenomena, by comparing the new likelihood value with the immediately preceding likelihood value;   compare the value representing the index of surprise with a threshold;   calculate, depending on the outcome of the comparison with the threshold, a new highest average sampling rate for the node in that state that will still result in the energy stored at the node being sufficient to continue operating the node for the user-specified lifetime; and,   set a new sampling rate for each sensor of the node, either above or below the new highest average sample rate, in proportion with the likelihood (PMF) of that node's current most likely state compared to the other states of the phenomena.   
     
     
         6 . A wireless sensor node according to  claim 5 , wherein the node has a range of different types of sensors. 
     
     
         7 . A wireless sensor node according to  claim 6 , wherein each cycle of operation of the computer system of a sensor node involves the additional step of setting a new sampling rate for each of the set of sensors associated with that sensor node, according to a predetermined regime for that state of the phenomena. 
     
     
         8 . A wireless sensor node according to  claim 1 , wherein a statistical model automatically learns about the nature of each phenomena and determines an optimal function for assigning sampling frequencies to the state of the phenomena. 
     
     
         9 . A wireless sensor node according to  claim 8 , wherein the statistical model monitors the rate of change of the data from the sensors to identify peaks of surprise. 
     
     
         10 . A wireless sensor node according to  claim 8 , wherein the automatic learning takes place on a computer at the network hub. 
     
     
         11 . A wireless sensor node according to  claim 10 , wherein once an optimal function is determined a corresponding algorithm may be downloaded to the node where it assigns new sampling frequencies as required. 
     
     
         12 . A network of sensor nodes comprising:
 a plurality of sensor nodes, each sensor configured to operate for a given lifetime, comprising:   an onboard computer system; and   a set of one or more associated sensors;   wherein, the computer system is configured to:
 periodically sample data from each sensor of the set of associated sensors, 
 store a multi-state model representing one or more phenomena described by the collected data; 
 calculate a value associated with movement of the phenomena between the states of the multi-state model; and 
 adjust the rate of sampling of one or more of the set of associated sensors depending on the calculated value; and 
   wherein each sensor node learns the minimum residual energy in its region of the network, and is able to set its sensor sampling rates in order to conserve sufficient energy at critical nodes of the network along its reporting path.   
     
     
         13 . A method of operating a sensor node comprising an onboard computer system and a set of one or more associated sensors, wherein the node is configured to operate for a given lifetime; the method comprising the steps of:
 periodically sampling data from a set of one or more sensors associated with a sensor node;   storing a multi-state model representing one or more phenomena described by the collected data;   calculating a value associated with movement of the phenomena between the states of the multi-state model; and,   adjusting the rate of sampling of one or more of the set of associated sensors depending on the calculated value.   
     
     
         14 . The method of  claim 13  further comprising:
 collecting a set of fresh data from the set of sensors; 
 using the fresh set of data to generate a new likelihood value for the most likely state the phenomena is in; 
 calculating a value representing an index of surprise associated with movements of the phenomena, by comparing the new likelihood value with the immediately preceding likelihood value; 
 comparing the value representing the index of surprise with a threshold; 
 calculating, depending on the outcome of the comparison with the threshold, a new highest average sampling rate for the node in that state that will still result in the energy stored at the node being sufficient to continue operating the node for the user-specified lifetime; and, 
 setting a new sampling rate for each sensor of the node, either above or below the new highest average sample rate, in proportion with the likelihood (PMF) of that node's current most likely state compared to the other states of the phenomena. 
 
     
     
         15 . A wireless sensor node according to  claim 2 , wherein the multi-state model also includes a probability mass function (PMF) for each state to describe a likelihood of a measurement returning a particular value while the phenomena is in that state. 
     
     
         16 . A wireless sensor node according to  claim 2 , wherein the multi-state model further includes a transition weight for each respective transition between states, defining the likelihood of each transition; or no transition. 
     
     
         17 . A wireless sensor node according to  claim 3 , wherein the multi-state model further includes a transition weight for each respective transition between states, defining the likelihood of each transition; or no transition. 
     
     
         18 . A wireless sensor node according to  claim 9 , wherein the automatic learning takes place on a computer at the network hub. 
     
     
         19 . A wireless sensor node according to  claim 2 , wherein-the computer system is configured, during each cycle of operation, to:
 collect a set of fresh data from the set of sensors;   use the fresh set of data to generate a new likelihood value for the most likely state the phenomena is in;   calculate a value representing an index of surprise associated with movements of the phenomena, by comparing the new likelihood value with the immediately preceding likelihood value;   compare the value representing the index of surprise with a threshold;   calculate, depending on the outcome of the comparison with the threshold, a new highest average sampling rate for the node in that state that will still result in the energy stored at the node being sufficient to continue operating the node for the user-specified lifetime; and,   set a new sampling rate for each sensor of the node, either above or below the new highest average sample rate, in proportion with the likelihood (PMF) of that node's current most likely state compared to the other states of the phenomena.   
     
     
         20 . A wireless sensor node according to  claim 3 , wherein the computer system is configured, during each cycle of operation, to:
 collect a set of fresh data from the set of sensors;   use the fresh set of data to generate a new likelihood value for the most likely state the phenomena is in;   calculate a value representing an index of surprise associated with movements of the phenomena, by comparing the new likelihood value with the immediately preceding likelihood value;   compare the value representing the index of surprise with a threshold;   calculate, depending on the outcome of the comparison with the threshold, a new highest average sampling rate for the node in that state that will still result in the energy stored at the node being sufficient to continue operating the node for the user-specified lifetime; and,   set a new sampling rate for each sensor of the node, either above or below the new highest average sample rate, in proportion with the likelihood (PMF) of that node's current most likely state compared to the other states of the phenomena.