Healthor: Heterogeneity-aware Flow Control in DLTs to Increase Performance and Decentralization

Healthor is a novel heterogeneity-aware, lightweight flow-control mechanism for permissionless reputation-based DLT networks. It captures heterogeneity by defining a node's health as a function of its processing power and the current network activity. The health updates of neighboring nodes are then used to calculate the message forwarding rates, thus dynamically adjusting the flow control per node. This basic mechanism allows high-end nodes to buffer messages for unhealthy nodes, thus protecting weaker nodes from being overwhelmed with wasted processing and rapidly adapting network load and bursts. With such a flow-control design more nodes are able to keep up with ledger updates and participate in DLT consensus, therefore increasing decentralization and network performances.

This dataset contains the raw simulation data as shown in the author's Master's thesis. Due to the size of the raw data the dataset is split into several parts: 

attacks.tbz2: Results of the attack analysis. 

macrobenchmarks.tbz2: Results of marcobenchmarks up to networks with 2000 nodes.

macrobenchmarks5000.tbz2: Results of macrobenchmarks with a network of 5000 nodes.

microbenchmarks.tbz2: Results of the microbenchmarks.

sensitivity-analysis.tbz2: Results of Healthor's sensitivity analysis.

Please note that the full unpacked archive is over 12.5 GB of data, made up of hundreds of thousands of small files. 

Details on the archive format
Every archive contains several different file types. All of them together are necessary to reproduce experiment results and make sense of the data overall. In general, all files are grouped together by their simulation name as a prefix (e.g. "ExceedSendingRateAttack*"). 

"[simulation name].csv" contains overall information about the network setup like processing and generation rates.

"[simulation name]-generated[node number].csv" contains all messages and their generation time of the node given by node number.

"[simulation name]-processed_[node number].csv" contains all messages and their processing time of the node given by node number.

Any ".sca", ".vci" and ".vec" files are produced by OMNeT++ and contain the information rendered out into "[simulation name].csv" but are only readable within the OMNeT++ toolset.

See https://github.com/jonastheis/healthor for detailed instructions how to use this data to reproduce results shown in the thesis using this dataset.