Edge-computing is one of the most promising techniques to leverage the excess capacity that exists at users' premises. Unfortunately, edge-computing may be vulnerable to free-riding, i.e., to nodes that attempt to benefit from the system without providing any service in return. Traditional approaches model free-riders as rational nodes that strive to maximize a utility and apply Game Theory concepts to devise mechanisms that deny any utility gain to nodes that deviate from the protocol. FastRank was proposed to devise lightweight mechanisms targeted for the optimistic setting where the vast majority of nodes adopts one of a small number of behaviours. More precisely, it assumes that most nodes are altruistic or follow non-sophisticated behaviours such as free-riding or white-washing. If a small fraction of nodes follows alternative behaviours, then fastRank lightweight mechanism limits the utility gain of these nodes, making it unlikely that the number of nodes exploiting sophisticated behaviours increases at a fast pace. This allows altruistic nodes to detect their presence in time to switch to more robust mechanisms, before the system reaches a state where the lightweight mechanisms can no longer cope with the existing behaviours.

FastRank is a peer-to-peer streaming service designed to be stable with a setting of several freeriders and altruist nodes, being freeriders the ones who don't contribute to the streaming and only that advantage of it. FastRank has an overlay network construction and maintenance protocol, a ranking mechanism and a dissemination mechanism. It also offers a more sophisticated solution by providing a adaptive framework which allows to change to a more sophisticated strategy when facing environments more complicated than the above mentioned. By the use of several metrics throughout the system the source can collect several information from many of the altruistic nodes present in the network thus being able to determine if a different and more robust strategy is necessary, if so the source broadcasts a message indicating the nodes to switch to a more robust protocol. The metrics used to measure the state of the network is the number of kicks a node can give to other freeRider nodes, afterwards, when several complains are made to the source, majority votes are counted and a weighted average of several complain rounds is done in order to understand if there is a necessity of change. Our current step is to make a real scale simulation of this switching protocol on the PlanetLab EU framework and measure it's effectiveness.

Evaluation in a realistic scenario will use a slice in PlanetLab, with each node representing one or more members of a Peer-to-Peer (P2P) network. Peers will exchange messages among themselves and with a host located in Faculdade de Ciências da Universidade de Lisboa (IP address 194.117.20.220) that will be used for coordination and bootstrap of the network. No unsolicited messages will be sent for nodes outside the slice.

Traffic will be mostly composed of messages exchanged among peers using TCP connections. Each peer is expected to send and receive 10 messages/s, for an overall bandwidth use that should not exceed 200Kbit/s. Assuming 5 peers per node, the estimated bandwidth for a single PlanetLab node is therefore of approximately 1Mbit/s.

To simulate free-riders ejection mechanisms, connections are not expected to be stable and therefore one should expect some occasional spikes in the number of connection requests.

The FastRank algorithms include mechanisms which may occasionally require CPU intensive operations, in particular, for the resolution of challenges in the form of crypto puzzles. This is a key aspect of the algorithm and is not expected to bring any negative impact in the overall performance of PlanetLab considering the existing isolation among nodes.

Gustavo Correia : fc41960@alunos.fc.ul.pt
Tiago Maurício : tiagolcmauricio@gmail.com
Hugo Miranda (Technical contact) : hamiranda@ciencias.ulisboa.pt
Luis Rodrigues : ler@tecnico.ulisboa.pt
Xavier Vilaça : xvilaca@gsd.inesc-id.pt