Wireless Sensor Network Positioning Techniques
Doctoral thesis, 2013
Position information is one of the key requirements for wireless sensor networks
(WSNs). Since GPS receivers have drawbacks when used in low-power sensor nodes,
mainly due to latency in position recovery and limited access to GPS satellites in indoor-
type scenarios, extracting the position information by means of the network itself, so-
called positioning or localization, has been considered an eective solution for the posi-
tioning problem. The goal of this thesis is to design and develop approaches, algorithms,
and benchmarks for the positioning problem for WSNs. The contributions of the thesis,
which appear in the appended papers, are as follows:
Paper A develops an eavesdropping technique for positioning multiple target nodes
in a cooperative wireless sensor network in the presence of unknown turnaround times.
That is, a number of reference or target nodes (secondary nodes) can listen to both sig-
nals transmitted by the target and an active (primary) node. The maximum likelihood
estimator (MLE) and a theoretical lower bound as well as a suboptimal ecient linear
estimator are derived for the problem. Numerical results conrm a considerable im-
provement for the proposed technique compared to conventional approaches, especially
for low signal-to-noise ratios. Paper B studies a self-positioning problem based on TDOA
measurements in the presence of unknown target node clock skew. Since the optimal
MLE poses a dicult global optimization problem, two suboptimal estimators followed
by a ne-tuning approach are investigated in this paper. Numerical results show that
the suboptimal estimators asymptotically attains the Cramer-Rao lower bound. Paper C
investigates the single target node localization problem based on received signal strength
measurements in the presence of unknown channel parameters. Using approximations,
the problem is rendered to a low complex problem and a simple technique is employed to
solve the problem. The proposed technique shows a good trade-o between accuracy and
complexity compared to the existing approaches. Paper D studies the possibility of up-
per bounding the position error for range-based positioning algorithms in wireless sensor
networks. It is argued that in certain situations when the measured distances between
sensor nodes have positive errors, the target node is conned to a closed bounded con-
vex set, which can be derived from the measurements. In particular, the upper bounds
are formulated as nonconvex optimization problems, and relaxation techniques are em-
ployed to approximately solve the nonconvex problems. Simulation results show that
the proposed bounds are reasonably tight in many situations, especially for non-line-of-
sight conditions. Finally, Paper E deals with identifying the feasible sets in cooperative
positioning and proposes an iterative technique to cooperatively outer-approximate the
feasible sets containing the locations of the target nodes. Simulation results show that
the proposed technique converges after a small number of iterations.
outer-approximations
Cram
Wireless sensor network
nonlinear and linear least squares
cooperative positioning
projection onto convex sets
maximum likelihood esti- mator