This demonstrated that designing an
energy e?±cient MAC protocol in a multi-hop setting is still an important
issue, even with low volumes of tra?±c.
2. Reliable Packet Delivery: As the sensory data was oversampled by a
factor of three, no link layer retransmissions were employed to guarantee
successful data delivery. Hence in the case of a single-hop network, the
packet delivery rate met expectations. However in the case of a multi-hop
network, the packet delivery rate was smaller than expected. The collisions
and the congestion was not a major factor, as the nodes used a CSMA
MAC protocol and the total network utilization was small. It was concluded
that the single-hop packet loss rate along with the correlated node
failures lead to substantial packet losses that increased exponentially with
the number of hops. This suggested that even a communication protocol
that employs only link layer retransmissions, but ignores wider network
outages, may not be su?±cient for reliable packet delivery. Hence end-toend
transfer approaches are necessary for reliable packet delivery.
3. Node Failures: During the first set of experiments in summer 2002, significant
packet collisions were observed for some nodes even though the
expected network utilization was very low. Some of the collisions could
have been triggered due to clock skew, however it was concluded that the
23 Chapter 1 Design of Large-scale Sensor Networks
Aravind Iyer et al.
Pages:
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75