Wireless Sensor Networks (WSNs) are a popular class of distributed collaborative networks finding suitability from medical to military applications. However, their vulnerability to capture, their "open" wireless interfaces, limited battery life, all result in potential vulnerabilities. WSN-based services inherit these vulnerabilities. We focus on tactical environments where sensor nodes play complex roles in data sensing, aggregation and decision making. Services in such environments demand a high level of reliability and robustness.

The first problem we studied is robust target localization. Location information is important for surveillance, monitoring, secure routing, intrusion detection, on-demand services etc. Target localization means tracing the path of moving entities through some known surveillance area. In a tactical environment, an adversary can often capture nodes and supply incorrect surveillance data to the system. In this thesis we create a target localization protocol that is robust against large amounts of such falsified data. Location estimates are generated by a Bayesian maximum-likelihood estimator. In order to achieve improved results with respect to fraudulent data attacks, we introduce various protection mechanisms. Further, our novel approach of employing watchdog nodes improves our ability to detect anomalies reducing the impact of an adversarial attack and limiting the amount of falsified data that gets accepted into the system. By concealing and altering the location where data is aggregated, we restrict the adversary to making probabilistic "guess" attacks at best, and increase robustness further. By formulating the problem of robust node localization under adversarial settings and casting it as a multivariate optimization problem, we solve for the system design parameters that correspond to the optimal solution. Together this results in a highly robust protocol design.

In order for any collaboration to succeed, collaborating entities must have the same relative sense of time. This ensures that any measurements, surveillance data, mission commands, etc will be processed in the same epoch they are intended to serve. In most cases, data disseminated in a WSN is transient in nature, and applies for a short period of time. New data routinely replaces old data. It is imperative that data be placed in its correct time context; therefore...