Image Processing Reference
In-Depth Information
The position information once obtained was not only used for characterizing the data, but also in
designing the networking protocols, for example, leading to more efficient routing schemes based on
the estimated position of the nodes [].
The second important building block is the timing and synchronization block. Nodes are allowed
to function in a sleep mode for long period, so periodic waking-up intervals need to be computed
within a certain precision. However, the notion of local time and synchronization with the neigh-
bors is needed for the communication protocols to perform well. Light-weight algorithms have been
developed that allow fast synchronization between neighboring nodes using a limited number of
messages.Loosesynchronizationistobeused,meaningthateachpairofneighbornodesaresynchro-
nized within a certain bound, while nodes situated multiple hops away might not be synchronized
at all.
Global timing notion might not be needed at all in most of the applications. Due to the fact that
many applications measure natural phenomenon such as temperature, where delays up to the order
of seconds can be tolerated, the trade-off between latency and energy is preferred.
The last important block is the signal processing unit. A new class of algorithms has to be developed
due to the distributed nature of wireless sensor networks. In their vast majority the signal processing
algorithms are centralized algorithms that require a large computation power and the availability of
all the data at the same time. Transmitting all the recorded data to all nodes is impossible in a dense
network even from theoretical point of view, not to mention the needed energy for such an operation.
The new distributed signal processing algorithms have to take into account the distributed nature of
the network, the possible unavailability of data from certain regions due to failures and the time
delays that might be involved.
3.4.4 Dependability
More than any other sort of computer network, the wireless sensor networks are subject to failures.
Unavailability of services are considered “a feature” of these networks or “regular events” rather than
some sporadic and highly improbable events. he probability for something going wrong is at least
several orders of magnitude higher than in all the other computer networks.
All the algorithms have to employ some form of robustness in front of the failures that might
affect them. On the other hand, this comes at the cost of energy, memory, and computation power,
so it has to be kept at a minimum. An interesting issue is the one of the system architecture from
the protocols point of view. In traditional computer networks, each protocol stack is designed for
the worst case scenario. This scenario hardly ever happens simultaneously for all the layers and a
combination of lower layer protocols could eliminate such a scenario. his leads to lot of redundancy
in the sensor node, redundancy that costs important resources. The preferred approach is that of
cross-layer designing and studying of the sensor node as a whole object rather than separate building
blocks. This opens for a discussion about the topic of what is a right architecture for all the sensor
networks and if a solution that fits all the scenarios makes sense at all.
Let us summarize the sources of errors the designer will be facing: nodes will stop functioning
starting with even the (rough) deployment phase. he harsh environment will continuously degrade
the performances of the nodes making them unavailable as the time passes. hen, the wireless com-
munication medium will be an important factor to disturb the message communication and to affect
the links and implicitly the network topology. Even with a perfect environment, collisions will occur
due to the imprecise local time estimates and lack of synchronization. Nevertheless, the probabilistic
scheduling policies and protocol implementations can be considered sources of errors.
Another issue that can be addressed as a dependability attribute is the security. he communication
channel is opened and cannot be protected. his means that others are able to intercept and to disrupt
the transmissions or even to transmit their own data. In addition to accessing private information, a
third party could also act as an attacker that wants to disrupt the correct functionality of the network.
