The Internet of Things (IoT) will result in the deployment of many billions of wireless embedded systems creating interactive pervasive environments. It is envisaged that devices will cooperate to provide greater system knowledge than the sum of its parts. In an emergency situation, the flow of data across the Internet of Things may be disrupted, giving rise to a requirement for machine-to-machine interaction within the remaining ubiquitous environment. Geographic Hash Tables (GHTs) provide an efficient mechanism to support fault-tolerant rendezvous communication between devices. However, current approaches either rely on devices being equipped with a GPS or being manually assigned an identity. This is unrealistic when the majority of these systems will be located inside buildings and will be too numerous to expect manual configuration. Additionally when using GHT as a distributed data store, imbalance in the topology can lead to storage and routing overhead. This causes unfair work load, exhausting limited power supplies as well as causing poor data redundancy. To deal with these issues we propose an approach that balances graph-based layout identity assignment, through the application of multi fitness genetic algorithms. Our experiments show through simulation that our multi fitness evolution technique improves on the initial graph-based layout, providing devices with improved balance and reachability metrics. Index Terms-data centric storage, evolutionary computing and genetic algorithms, information dispersal, load balancing, Wireless sensor networks