Query Processing in Sensor Networks 
=====================================

Advantages:
- By adopting in-network processing reduces energy consumption and reduce bandwidth usage
- By using gateway nodes the energy needed by the sensor nodes to transmit data to 
  the base station is conserved.
- The sensor data can be cached in the leader node or the gateway node that has more 
  processing and capabilities. This is suitable for applications that do not require 
  real time data. The stored data can be sent to the base station at certain intervals.
- The use of AODV as the routing protocol, the authors ensure that their proposed algorithm scales to
  large size networks. The fact that it does not generate duplicate data packets, 
  a requirement to do innetwork aggregation

Disadvantages:
- The authors suggest a template for the query, but leave out the design of the of a suitable query
  language.
- The proposed query template has only a limited usage for event-oriented applications.
- Messages are circulated periodically to check that the leader is alive increasing the energy overhead.
- The authors assume bi-directional links of same power which might not generally be the case in the sensor networks.

Improvements:
- I think for efficient query processing algorithms, there is a need to distinguish between the long
running queries: which deal with the status of the sensor network over a user defined time period, and
snapshot queries: which deal with the current status of the sensor network. Different strategies are
needed to handle the two types of queries in order to ensure efficient energy consumption.
- One can adopt a power based optimization technique for query processing. Here we can have the base
station optimize the query before dissemination. These include cost-based optimization that should
yield lowest overall power consumption. Then an optimizer can focus on ordering joins, selections and
sampling on individual nodes.
- We could form clusters based on the type of data that the nodes would generate.