This document provides an overview of IOTA, a distributed ledger technology designed for the Internet of Things. It summarizes IOTA's Tangle network which uses a directed acyclic graph structure instead of a blockchain. Key points include: IOTA aims to enable micropayments and data transfer between devices; the Tangle reaches consensus in a self-regulating way without fees by having nodes validate two previous transactions; and the tip selection algorithm uses a Markov chain Monte Carlo process to determine the validity of transactions and prevent double-spending attacks.
11. ● Every party has an
individual copy
● Prone to tampering
● High cost to compare
datasets
● No way of verifying the
original dataset
Traditional Ledgers
11
11
12. ● Datasets shared amongst all
parties
● Every party can verify the
datasets of other participants in
the network
● Tampered datasets are excluded
● An immutable single source of
truth is established
Distributed Ledger
12
19. ● Limited to Block Size and
Time
● Not scalable
● Transaction Censorship (fee
thresholds)
● Centralization
● Consensus Decoupled
● 80% of miners from China
● ...
“Include my
transaction, here is
my fee”
Blockchain’s Limitations
19
21. Tangle
● Bundles all transactions in a Directed Acyclic
Graph (DAG)
● Completely self-regulating, consensus no longer
decoupled
● Very Scalable. Low overhead Proof-of-Work to
prevent spam
21
32. Cloud, Fog and Mist
EDGECORE
● Real Time Processing
● Avoid network congestion
and signal collisions
● Incentive for interoperability
● Data Integrity
Requirements
32
34. Protocol and Network Statistics
◉ Official launch July 11th
◉ 1 year of development
and testing
◉ Community of 1000+
◉ More than 2m transactions
◉ $60m+ of value transferred
◉ More transactions per
second than any blockchain
34
35. (Unofficial slide)
◉ Tech problems with auto-discovery/slow syncing
and double-spending wallets
◉ Network currently down
◉ Community Management
◉ Funding of Foundation and Big Deal
◉ Anniversary update, Release 1.1, +1 Dev,
+1Bussines, Phase 2 (utility), Exchanges
36. IOTA Foundation
We are an open-source, non-profit Software
Foundation. Goal of the Foundation is it to
establish IOTA as a standard in the IoT stack.
46. Basics
● DAG Basic Examples (trees, direct, indirect, confirmation)
● Approvment DAG (IOTA) vs Transaction DAG (BTC)
● Main Idea: Issuing a Tx needs PoW (Approving other Tx)
● Approving means
Signature verification &
Check for double spendings
● But which Tx to approve???
47. A bit of terminology
● Genesis & Tips
● (own) weight (∝ amount of PoW)
● cumulative weight ● Score
48. Tip Selection Algorithm
● Only guideline for network, no rule
● Simple strategy: Choose two random tips
● Does not encourage approving tips by lazy nodes
● Because probability that their tips will get approved is equal to
the one of the tips that did real PoW
● Solution: Bias towards end (i.e. tips with high score)
49. Tip Selection Algorithm
● Simple strategy 2: Choose two tips out of the 10% with the
highest scores
● Problem: Vulnerable against double-spending “large weight”-
attack will eventually always succeed
50. Tip Selection Algorithm
● Mitigation strategy: Cap Max weight of transaction
● This drastically decreases probability of attacker
● But he can still beat us by pre-building a parasite-chain
51. A parasite
● Main tangle tip score ≈ Sum of main tangle weights
● Parasite tangle tip score ≈ Sum of main tangle and parasite
chain weights
● Parasite chain can be build without broadcasting
● Idea: main tangle should have more active
hashing power
● which is ∝ cum weight
52. The Final Tip Selection Algorithm
● How do we figure out which tip is on the main tangle?
● We place some „particles“ deep in the tangle
● And let them walk towards the tips
● With affinity to high cumulative weights (MCMC)
● This will very unlikely select lazy tips
● and very unlikely parasite chain tips
53. Double Spending
● Conflicting Tx can exist in the Tangle
● Network has to decide which Tx (SubTangle) will not get
orphaned
● Methodology: Run tip selection Algorithm many times and
see how often the conflicting Tx are (indirectly) approved by
the tips
● Result: Transactions in the subTangle with less hashing power
will eventually get orphaned
54. Interesting Details
● All addresses can only get used once
● Spamming the network actually helps it
● Coordinator helps network until enough nodes are online
● Tx actually have a particular structure (bundles)