Reina et al. propose a novel algorithm, Jaccard distance, to improve redundancy issues during the mobile ad-hoc network (MANET) discovery phase. The Jaccard distance algorithm is a probabilistic-based scheme that uses a single variable of neighbor distance of each node to calculate the Jaccard distance of nodes. The authors also prove that the Jaccard distance is correlated to the Euclidean distance. The significant difference between their new scheme and the existing one is that their scheme does not use global positioning system (GPS) information.

The introduction and related works sections are very well written. The background information on MANETs and existing schemes introduces the basis for the authors’ work and explains why it is different from existing approaches. Deterministic approaches may have disadvantages including packet flooding and energy inefficiency, while current probabilistic approaches may have some disadvantages including the need for GPS information, large delays, packet redundancy issues, and complexity. Therefore, it is critical to design a scheme that is load balanced, resilient, and simple.

The Jaccard distance algorithm itself is straightforward. The authors provide a good illustration of the algorithm with an example of networking conditions between two nodes. The relationship of two nodes in a MANET is divided into four areas: a1, a2, a3, and a4. By using graphic calculations, the Jaccard distance is formed. The question is: How fast can the algorithm be performed on each node? Especially considering a rapidly changing environment such as a MANET, all of the variables (a1 to a4) are highly dependent on current networking conditions. What happens when some of the neighbors quickly move out of the area and some nodes quickly move into the area? These kinds of questions should be factored into the algorithm. Another question: How much overhead is necessary to perform this kind of calculation on each node? Will this kind of algorithm be more energy efficient than current ones? The complexity of the algorithm and the computational overhead should be considered, too.

The authors also state that the Jaccard distance is correlated to the Euclidean distance. Through extensive simulation of various networking conditions, Jaccard distance-based mechanisms are compared with some other notable schemes. The results seem to be promising. The Jaccard distance-based mechanisms outperform other schemes in terms of normalized routing load (NRL) and they do not require GPS.