The ongoing Smart Grid (SG) initiative proposes several modications to the existing
power grid in order to better manage power demands, reduce CO2 emissions and
ensure reliability through several new applications. One part of the SG initiative that
is currently being implemented is the Advanced Metering Infrastructure (AMI) which
provides two-way communication between the utility company and the consumers'
smart meters (SMs).
The AMI can be built by using a wireless mesh network which enables multi-hop
communication of SMs. The AMI network enables collection of ne-grained power
consumption data at frequent intervals. Such a ne-grained level poses several privacy
concerns for the consumers. Eavesdroppers can capture data packets and analyze
them by means of load monitoring techniques to make inferences about household activities.
To prevent this, in this dissertation, we proposed several privacy-preserving
protocols for the IEEE 802.11s-based AMI network, which are based on data obfuscation,
fully homomorphic encryption and secure multiparty computation. Simulation
results have shown that the performance of the protocols degrades as the network
grows. To overcome this problem, we presented a scalable simulation framework for
the evaluation of IEEE 802.11s-based AMI applications. We proposed several modi-
cations and parameter adjustments for the network protocols being used. In addition,we integrated the Constrained Application Protocol (CoAP) into the protocol stack
and proposed ve novel retransmission timeout calculation functions for the CoAP
in order to increase its reliability.
Upon work showing that there are inconsistencies between the simulator and a
testbed, we built an IEEE 802.11s- and ZigBee-based AMI testbed and measured
the performance of the proposed protocols under various conditions. The testbed is
accessible to the educator and researchers for the experimentation.
Finally, we addressed the problem of updating SMs remotely to keep the AMI
network up-to-date. To this end, we developed two secure and reliable multicast-overbroadcast
protocols by making use of ciphertext-policy attribute based signcryption
and random linear network coding. |