Abstract:Abstract-This paper presents two new greedy sensor placement algorithms, named minimum nonzero eigenvalue pursuit (MNEP) and maximal projection on minimum eigenspace (MPME), for linear inverse problems, with greater emphasis on the MPME algorithm for performance comparison with existing approaches. In both MNEP and MPME, we select the sensing locations one-by-one. In this way, the least number of required sensor nodes can be determined by checking whether the estimation accuracy is satisfied after each sensing… Show more
“…These bounds are also reflected in the results from Fig. 2 where we can observe that for k ≈ n the decrease in MSE achieved by the proposed method, with the increased number of selected sensors k, is larger that of a random sensor selection algorithm when k n. These insights confirm previous experimental results from the literature, like [4], where the methods proposed for sensors selection differ mostly when k ≈ n and are similar when k n (or k ≈ m) where even random selections provide good estimation accuracy (low MSE and WCE).…”
Section: B Relating Sensor Management To Other Problemssupporting
confidence: 88%
“…2: the estimated curve is above the one empirically observed. It is clear from the figure that the largest differences are for low k (on the same order with n) while the gap closes for k approaching m. As we will also see from the results section, the largest differences between the performance of the methods we analyze are for low values of k. Indeed, past research [4] has shown by numerical experimentation that most of the sensor selection methods proposed in the literature perform similarly in the regime k n. Also, Result 4 shows that the MSE(A 1 ) and WCE(A 1 ) decrease on average linearly with the number of selected sensors. Dependencies with the other dimensions are also linear and intuitive: increasing the number of parameters to estimate (n) and the total number of available sensors (m) leads to worse performance; increasing the energy, essentially the signal to noise ratio, of the measurement matrix (α) improves performance.…”
Section: A Results For a Tight Sensor Networkmentioning
confidence: 53%
“…A proof is given in [39,Chapter 4]. This is one of the reasons given in [4] that the goal to increase all the eigenvalues λ i with each new measurement. Consider also the following two results.…”
Section: B Greedy Methods Approachmentioning
confidence: 99%
“…Fig. 4 shows the simulation results where we compare with FrameSense [6], convex relaxation ( 1 followed by rounding, using the log determinant approach to minimize VCE) [8], SparSenSe [13] and MPME [4]. All measurement matrices used here are α−tight with α = 100.…”
Section: B Comparisons With Previous Sensor Selection Algorithmsmentioning
confidence: 99%
“…The sensor placement (and in general the sensor management) problems have been extensively studied in the past. A general approach is to use greedy methods based on a minimum eigenspace approach [4] or with submodularity based performance guarantees [5] that provide results within (1 − e −1 ) of the optimal solution. Another popular greedy sensor selection method, called FrameSense [6], initially activates all the sensors and then removes one at each step based on a "worst-out" principle to optimize its submodular objective function.…”
In this paper we present new algorithms and analysis for the linear inverse sensor placement and scheduling problems over multiple time instances with power and communications constraints. The proposed algorithms, which deal directly with minimizing the mean squared error (MSE), are based on the convex relaxation approach to address the binary optimization scheduling problems that are formulated in sensor network scenarios. We propose to balance the energy and communications demands of operating a network of sensors over time while we still guarantee a minimum level of estimation accuracy. We measure this accuracy by the MSE for which we provide average case and lower bounds analyses that hold in general, irrespective of the scheduling algorithm used. We show experimentally how the proposed algorithms perform against state-of-the-art methods previously described in the literature.
“…These bounds are also reflected in the results from Fig. 2 where we can observe that for k ≈ n the decrease in MSE achieved by the proposed method, with the increased number of selected sensors k, is larger that of a random sensor selection algorithm when k n. These insights confirm previous experimental results from the literature, like [4], where the methods proposed for sensors selection differ mostly when k ≈ n and are similar when k n (or k ≈ m) where even random selections provide good estimation accuracy (low MSE and WCE).…”
Section: B Relating Sensor Management To Other Problemssupporting
confidence: 88%
“…2: the estimated curve is above the one empirically observed. It is clear from the figure that the largest differences are for low k (on the same order with n) while the gap closes for k approaching m. As we will also see from the results section, the largest differences between the performance of the methods we analyze are for low values of k. Indeed, past research [4] has shown by numerical experimentation that most of the sensor selection methods proposed in the literature perform similarly in the regime k n. Also, Result 4 shows that the MSE(A 1 ) and WCE(A 1 ) decrease on average linearly with the number of selected sensors. Dependencies with the other dimensions are also linear and intuitive: increasing the number of parameters to estimate (n) and the total number of available sensors (m) leads to worse performance; increasing the energy, essentially the signal to noise ratio, of the measurement matrix (α) improves performance.…”
Section: A Results For a Tight Sensor Networkmentioning
confidence: 53%
“…A proof is given in [39,Chapter 4]. This is one of the reasons given in [4] that the goal to increase all the eigenvalues λ i with each new measurement. Consider also the following two results.…”
Section: B Greedy Methods Approachmentioning
confidence: 99%
“…Fig. 4 shows the simulation results where we compare with FrameSense [6], convex relaxation ( 1 followed by rounding, using the log determinant approach to minimize VCE) [8], SparSenSe [13] and MPME [4]. All measurement matrices used here are α−tight with α = 100.…”
Section: B Comparisons With Previous Sensor Selection Algorithmsmentioning
confidence: 99%
“…The sensor placement (and in general the sensor management) problems have been extensively studied in the past. A general approach is to use greedy methods based on a minimum eigenspace approach [4] or with submodularity based performance guarantees [5] that provide results within (1 − e −1 ) of the optimal solution. Another popular greedy sensor selection method, called FrameSense [6], initially activates all the sensors and then removes one at each step based on a "worst-out" principle to optimize its submodular objective function.…”
In this paper we present new algorithms and analysis for the linear inverse sensor placement and scheduling problems over multiple time instances with power and communications constraints. The proposed algorithms, which deal directly with minimizing the mean squared error (MSE), are based on the convex relaxation approach to address the binary optimization scheduling problems that are formulated in sensor network scenarios. We propose to balance the energy and communications demands of operating a network of sensors over time while we still guarantee a minimum level of estimation accuracy. We measure this accuracy by the MSE for which we provide average case and lower bounds analyses that hold in general, irrespective of the scheduling algorithm used. We show experimentally how the proposed algorithms perform against state-of-the-art methods previously described in the literature.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.