Limits...
Relocatable, automated cost-benefit analysis for marine sensor network design.

D'Este C, de Souza P, Sharman C, Allen S - Sensors (Basel) (2012)

Bottom Line: We describe a novel automated method for generating and combining cost and benefit values to decide on the best sensor locations using information about the specific constraints available in most coastal locations.Benefits in maximum coverage and reducing overall error are also determined using model output.This method demonstrates equivalent accuracy at predicting the whole system to expert-chosen locations, whilst significantly reducing the estimated costs.

View Article: PubMed Central - PubMed

Affiliation: Tasmanian ICT Centre, CSIRO, Castray Esplanade, Hobart, TAS 7000, Australia. Claire.DEste@csiro.au

ABSTRACT
When designing sensor networks, we need to ensure they produce representative and relevant data, but this must be offset by the financial cost of placing sensors. We describe a novel automated method for generating and combining cost and benefit values to decide on the best sensor locations using information about the specific constraints available in most coastal locations. Costs in maintenance, negotiation, equipment, exposure and communication are estimated using hydrodynamic models and Electronic Navigation Charts. Benefits in maximum coverage and reducing overall error are also determined using model output. This method demonstrates equivalent accuracy at predicting the whole system to expert-chosen locations, whilst significantly reducing the estimated costs.

No MeSH data available.


Related in: MedlinePlus

Locations of 5 nodes in Moreton Bay handpicked by a hydrodynamic model expert (left) and chosen using the ACBA method (right).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3376612&req=5

f20-sensors-12-02874: Locations of 5 nodes in Moreton Bay handpicked by a hydrodynamic model expert (left) and chosen using the ACBA method (right).

Mentions: We asked a hydrodynamic modeller to pick some locations by hand. Figure 20 shows the expert-chosen locations and those chosen by ACBA. The locations are very similar, however the expert chosen locations have a combined cost score of 8.7 (average 1.74) and the ACBA locations have a combined score of 5.9 (average 1.18). The average error was calculated using a separate test set of 100 examples over the entire model area interpolated from the five chosen locations. The average error for the expert-chosen locations is 3.0% and the ACBA locations is 3.2%.


Relocatable, automated cost-benefit analysis for marine sensor network design.

D'Este C, de Souza P, Sharman C, Allen S - Sensors (Basel) (2012)

Locations of 5 nodes in Moreton Bay handpicked by a hydrodynamic model expert (left) and chosen using the ACBA method (right).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3376612&req=5

f20-sensors-12-02874: Locations of 5 nodes in Moreton Bay handpicked by a hydrodynamic model expert (left) and chosen using the ACBA method (right).
Mentions: We asked a hydrodynamic modeller to pick some locations by hand. Figure 20 shows the expert-chosen locations and those chosen by ACBA. The locations are very similar, however the expert chosen locations have a combined cost score of 8.7 (average 1.74) and the ACBA locations have a combined score of 5.9 (average 1.18). The average error was calculated using a separate test set of 100 examples over the entire model area interpolated from the five chosen locations. The average error for the expert-chosen locations is 3.0% and the ACBA locations is 3.2%.

Bottom Line: We describe a novel automated method for generating and combining cost and benefit values to decide on the best sensor locations using information about the specific constraints available in most coastal locations.Benefits in maximum coverage and reducing overall error are also determined using model output.This method demonstrates equivalent accuracy at predicting the whole system to expert-chosen locations, whilst significantly reducing the estimated costs.

View Article: PubMed Central - PubMed

Affiliation: Tasmanian ICT Centre, CSIRO, Castray Esplanade, Hobart, TAS 7000, Australia. Claire.DEste@csiro.au

ABSTRACT
When designing sensor networks, we need to ensure they produce representative and relevant data, but this must be offset by the financial cost of placing sensors. We describe a novel automated method for generating and combining cost and benefit values to decide on the best sensor locations using information about the specific constraints available in most coastal locations. Costs in maintenance, negotiation, equipment, exposure and communication are estimated using hydrodynamic models and Electronic Navigation Charts. Benefits in maximum coverage and reducing overall error are also determined using model output. This method demonstrates equivalent accuracy at predicting the whole system to expert-chosen locations, whilst significantly reducing the estimated costs.

No MeSH data available.


Related in: MedlinePlus