|
The metrics of quality of service (QoS) for each sensor type in a wireless sensor network can be associated with metrics
for multimedia that describe the quality of fused information, e.g., throughput, delay, jitter, packet error rate,
information correlation, etc. These QoS metrics are typically set at the highest, or application, layer of the protocol
stack to ensure that performance requirements for each type of sensor data are satisfied. Application-layer metrics, in
turn, depend on the support of the lower protocol layers: session, transport, network, data link (MAC), and physical.
The dependencies of the QoS metrics on the performance of the higher layers of the Open System Interconnection (OSI)
reference model of the WSN protocol, together with that of the lower three layers, are the basis for a comprehensive
approach to QoS optimization for multiple sensor types in a general WSN model. The cross-layer design accounts for
the distributed power consumption along energy-constrained routes and their constituent nodes. Following the author's
previous work, the cross-layer interactions in the WSN protocol are represented by a set of concatenated protocol
parameters and enabling resource levels. The "best" cross-layer designs to achieve optimal QoS are established by
applying the general theory of martingale representations to the parameterized multivariate point processes (MVPPs) for
discrete random events occurring in the WSN. Adaptive control of network behavior through the cross-layer design is
realized through the parametric factorization of the stochastic conditional rates of the MVPPs. The cross-layer protocol
parameters for optimal QoS are determined in terms of solutions to stochastic dynamic programming conditions derived
from models of transient flows for heterogeneous sensor data and aggregate information over a finite time horizon.
Markov state processes, embedded within the complex combinatorial history of WSN events, are more computationally
tractable and lead to simplifications for any simulated or analytical performance evaluations of the cross-layer designs.
|
