Subnet deduplication: fix interpretation of price premium

This commit is contained in:
Rucknium 2025-02-14 21:18:07 +00:00
parent 3fe27b56a5
commit 7e4a3c9625
2 changed files with 7 additions and 7 deletions

View file

@ -90,9 +90,9 @@
\begin{document}
\title{Subnet Deduplication for Monero Node Peer Selection\\\vspace{.3cm}
\large Draft v0.1\vspace{-.715cm}}
\large Draft v0.2\vspace{-.715cm}}
\author{Rucknium\orcidlink{0000-0001-5999-8950} }
\date{February 12, 2025}
\date{February 14, 2025}
\maketitle
\begin{abstract}
Spying adversaries can set up nodes on the Monero network to try to
@ -472,19 +472,19 @@ $p_{s,s}>p_{d,d}$ if and only if $w_{s}h_{s}<w_{d}h_{d}$. Rearranging,
we have this condition:
\begin{equation}
\dfrac{w_{s}}{w_{d}}<\dfrac{h_{d}}{h_{s}}
\dfrac{w_{d}}{w_{s}}>\dfrac{h_{s}}{h_{d}}
\end{equation}
This inequality says that subnet deduplication is a better strategy
for the honest node if the price premium of leasing subnet-distinct
IP addresses is less than the ratio of the number of distinct subnets
with at least one honest node to the total number of honest nodes.
IP addresses is more than the ratio of the total number of honest
nodes to the number of distinct subnets with at least one honest node.
Note that this condition does not depend on the adversary's budget.
At any given moment, $h_{d}/h_{s}$ can be computed by performing
At any given moment, $h_{s}/h_{d}$ can be computed by performing
a network scan, assuming we can determine which nodes are honest.
Using the network scan and list of suspected spy nodes from the previous
section, we have $h_{d}/h_{s}=1.38$. That means that the subnet deduplication
section, we have $h_{s}/h_{d}=1.38$. That means that the subnet deduplication
algorithm is better than the status quo if the price premium to lease
subnet-distinct IP addreses is 38 percent or greater. Of course, the
subnet concentration of honest nodes can change over time.