A Comparative Study of Modified BBO Variants and Other Metaheuristics for Optimal Power Allocation in Wireless Sensor Networks - Advances in Heuristic Signal Processing and Applications - page 108

Ta b l e 5 . 5 (Continued)

Sensors Analytical CBBO CBBO-DE 2-Stage-CBBO-DE CDE CGA CPSO

ρ = 0 . 5

S1 1.1575 1.1331 1.1346 1.1309 1.0956 1.1332

S2 0.8641 0.8202 0.8218 0.8375 0.7977 0.8334

S3 0.7925 0.8655 0.8579 0.8633 0.8290 0.8456

S4 0.7110 0.1630 0.2273 0.1443 0.0462 0.2466

S5 0.6426 0.6246 0.6178 0.6078 0.8519 0.5773

S6 0.4774 0.4436 0.4932 0.4826 0.2903 0.5087

S7 8.99E-09 0.7289 0.6948 0.7138 0.7070 0.7152

S8 1.53E-09 0 0.0093 0.0057 0.0027 2.32E-05

S9 4.51E-09 4.98E-04 0.0062 0.0064 0.0874 6.46E-04

S10 3.95E-09 0 0.0172 0 0.1389 0.0018

Poptimal 3.8610 3.8580 3.8583 3.8584 3.8630 3.8582

5.6 Conclusion

The present work has considered the problem of optimal power scheduling for the
decentralized detection of a deterministic signal in a WSN with power and band-
width constrained distributed nodes. An efficient optimal power allocation scheme
has the potential of suitably turning off the nodes with poor channels and providing
significant system power savings. In this work, three variants of the BBO algorithm
have been proposed for the optimal power allocation in WSNs. These algorithms
have been compared with three other competing algorithms, i.e., three separately
developed constrained versions of the DE, GA, and PSO algorithms. It has been
shown that the CBBO-DE algorithm has outperformed the other competing algo-
rithms for several types of simulation case studies, including both independent local
observation cases and correlated observation cases. It has also been observed that,
in the case of a large number of sensors, CBBO emerged as the best performer.
Finally, in this work, the fusion center is given the information on channel condi-
tion, assumed to obey Rayleigh fading. For the situation where the channel condition
is changing rapidly, the power allocation needs to be updated dynamically to ensure
an optimum performance. However, if the changes are reasonably slow, there will
be enough time to properly update the power allocation. It can be argued that for ap-
plications operating in a dynamic environment, there are other potential alternatives
using methods for dynamic optimization.

References

1. Abadir, K., Magnus, J.: Matrix Algebra. Econometric Exercises 1. Cambridge University
Press, Cambridge (2005)
2. Akyildiz, I.F., Su, W., Sankasubramaniam, Y., Cayirci, E.: Wireless sensor networks: a survey.
Comput. Netw. 38 , 393-422 (2002)