Sub-optimal
performance since
only a portion of
single hop
neighbors are
considered.
All available information
is used to
complete the message
path.
The search window
limits which of the
neighboring nodes
will be considered.
Computation
Complexity
P points per hop:
P function
evaluations.
M nodes in the
neighborhood.
P distance
calculations :
P ?¤M?¤(2MULT +
1ADD/SUB).
Define the search
window per hop:
16ADD/SUB +
17MULT.
N nodes in the
search window.
N distance
calculations :
N ?¤ (2MULT +
1ADD/SUB).
Depending on how
the next hop is
chosen, P can be
small.
N will be smaller
than P*M. The
search window is
only calculated
once and is comparable
to P function
evaluations.
??? TBF requires the source to define the path as a continuous function in
t, while in TER the path is defined, regardless of complexity, at the sink
and distributed as a set of discrete points to the nodes. TBF may allow
for a di?®erent path to be used for every packet. This comes at the cost of
overhead and lack of support for complex curves.
??? TBF requires that every node discretizes the curve within its neighborhood.
The number of calculations is dependent on dt and the complexity
of the curve function. In TER, the curve is discretized a priori. The curve
only needs to be discritized once in either case, but discretizing it at the
sink saves power and allows for more complex curves.
??? TBF requires floating point calculations to discretize the curve.
Pages:
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609