Ethereum 2.0 is a major upgrade to improve the performance of Ethereum. It will increase transaction processing capacity and alleviate the problem of fees, which have been rising to the point where millions of dollars are spent every day. In this talk, I explain how Ethereum 2.0 can be secured as the next generation of decentralized application platforms and present the research we are working on as part of that effort. First, I talk about technologies to improve security, including client diversity, fuzzing using multiple clients, and the fee market protocol EIP-1559. Second, I introduce Shargri-La, a protocol development support software we are developing (*), a simulator that helps researchers and developers quickly test protocol hypotheses to improve the performance and security of Ethereum 2.0. It simulates state transitions at the transaction level of granularity in "sharding," a technique that divides the blockchain into multiple pieces. Finally, I present the results of multi-agent simulations under EIP-1559 by modeling users' selfish behaviors and show possible future problems and mitigation solutions.

1. Ethereum 2.0 Security
Naoya Okanami

2. Who Am I?
● Naoya Okanami, also known as minaminao.
● LayerX Labs Researcher, 2019-.
● Master's Program, University of Tsukuba, 2020-.
● IPA MITOU Program 2020
● CTF Team: ./Vespiary

3. Agenda
● Ethereum 2.0 Overview
● Ethereum 2.0 Security
● Shargri-La Overview
● Analysis of EIP-1559 and ETH Transfers
● Future Work of Shargri-La

4. Ethereum 2.0 Overview

5. ● Factor: Decentralized
Finance (DeFi)
● Limits the applications
● Bad for UX and security
● Increasing throughput is
required.
Today’s Fee

6. What is Ethereum 2.0?
A major upgrade to the Ethereum network,
improving scalablity and security
● Phase 0: Staking
○ Scheduled to launch in November!
● Phase 1: Sharding (saving data only)
● Phase 1.5: Move the Eth1 chain into an Eth2 shard chain
● Phase 2: Sharding (state execution)

7. Sharding
Split into multiple blockchains for improving its performance
Shard 1
Shard 2
Shard 3

8. Beacon Chain
provides random
numbers
Shard Chain
provides data
Shard 64
B1 B2 B3 B4 B5
VM
provides state
execution result
Shard 1
Main Chain
provides staking
B1
state
root
B2
state
root
B3
state
root
B4
state
root
B5
state
root

9. Ethereum 2.0 Security

10. Client Diversity
Ethereum 2.0 Clients
● Prysm (Go)
● Lighthouse (Rust)
● Nimbus (Nim)
● Lodestar (TypeScript)
● Teku (Java)
If one client is down, the others survive.

11. Client Diversity
https://medium.com/prysmatic-labs/eth2-medalla-testnet-incident-f7fbc3cc934a

12. beacon-fuzz: A differential fuzzer for Ethereum 2.0
https://github.com/sigp/beacon-fuzz

13. EIP-1559: Fee market change
● An improved scheme for the first price auction currently
used by Ethereum (same for Bitcoin)
● EIP-1559 introduces a dynamic minimum gas prce,
“basefee”.
● The basefee is adjusted to transaction demand.
● Highly likely to be introduced in Ethereum 2.0

14. EIP-1559: Fee market change
Strategic issues in 1559

15. EIP-1559: Fee market change
Strategic issues in 1559

16. Shargri-La
A Transaction-level Sharded Blockchain Simulator

17. What is Shargri-La?
A Transaction-level Sharded Blockchain Simulator
Shard 1
Shard 2
Shard 3

18. Simulation Targets
Transaction-level
P2P
Consensus
User Behaviour
Smart Contracts
Virtual Machine

19. ● Is the protocol working as it's supposed to?
● Isn't the protocol working because of the user's
behavior?
● Isn't there some kind of attack method?
● Which algorithm is more suitable?
● What are the optimal parameters?
● ...
● Doesn't cross-shard TX make the UX worse?
● How many shards should a user watch?
● What happens if the load is concentrated on a
particular shard?
Issues that can be solved with Shargri-La
Solved!!

20. Current Progress: Version 0.1.0 Released!
https://github.com/shargri-la/shargri-la

21. Current Progress: Version 0.1.0 Released!
https://ethresear.ch/t/shargri-la-a-transaction-level-sharded-blockchain-simulator/7936

22. Current Progress: Version 0.1.0 Released!
● The core of the simulator is
written in Rust
● The graph visualization is
written in Python
(Matplotlib, Pandas)

23. Version 0.1.0
Analysis of EIP-1559 and ETH Transfers

24. Overview

25. UserGraph: A Transaction Demand Model

26. Simulation Settings
● The number of shards: 64
● The number of users: 10,000
● A user will move the shard every 100 slots.
● UserGraph parameters are set by uniform random numbers.

27. ETH Transfers
● We assume that all the on-chain activities are only the
transfers of ETH.
● We didn't implement virtual machine that runs smart contracts
to avoid increasing complexity.

28. Cost-reducing Wallet
● The hypothesis that we have had since the beginning of our
research: "Users periodically change the shards they use to
avoid the increased costs of cross-shard transactions.”
● We assume that this user behavior is accomplished by the
user’s "wallet".

29. Experiment 1: No user switches shards.

30. Experiment 2: A minority of users switch shards with the minimum selection.

31. Experiment 2: A minority of users switch shards with the minimum selection.

32. Experiment 2: A minority of users switch shards with the minimum selection.

33. Experiment 2: A minority of users switch shards with the minimum selection.

34. Experiment 2: A minority of users switch shards with the minimum selection.

35. Experiment 3: A majority of users switch shards with the minimum selection.

36. Experiment 3: A majority of users switch shards with the minimum selection.

37. Experiment 3: A majority of users switch shards with the minimum selection.

38. Experiment 3: A majority of users switch shards with the minimum selection.

39. Experiment 3: A majority of users switch shards with the minimum selection.
● Users rush to the minimum fee shard.
● Stuck transactions occurs.
● It's not a good algorithm (and phenomenon) for users.

40. Experiment 4: A majority of users switch shards with weighted random selection.

41. Experiment 4: A majority of users switch shards with weighted random selection.

42. Experiment 4: A majority of users switch shards with weighted random selection.

43. Experiment 4: A majority of users switch shards with weighted random selection.

44. Experiment 4: A majority of users switch shards with weighted random selection.

45. Experiment 4: A majority of users switch shards with weighted random selection.
● Users rush to the minimum fee shard.
● Stuck transactions occurs.
● It's not a good algorithm for users.

46. Experiment 5: Switchers with the minimum selection, switchers with the weighted
random selection, and non-switchers co-exist.

47. Experiment 6: An extremely popular user exists.

48. Experiment 6: An extremely popular user exists.

49. Future Work of Shargri-La

50. Phase 1/1.5 Data Fee Market
● Ethereum 2.0 Phase 0 is scheduled to launch in November.
● Phase 1 (sharding as a data layer) research is required.
● A EIP-1559-like data fee market is supposed for Phase 1.

51. Q&A

52. Appendix

53. Comparison With Similar Systems
Simulation Targets Architecture
P2P
(Layer 0)
Consensus
(Layer 1)
User, TX
(Layer 1.5+)
Sharding
shardSim ◯ ◯ ☓ ◯
SimBlock ◯ ◯ ☓ ☓
VIBES ◯ ◯ ☓ ☓
Bitcoin-Simulator
(ns-3 based) ◯ ◯ ☓ ☓
Shargri-La ☓ ☓ ◯ ◯

54. Blockchain
･･･ ･･･

55. Scalability
● Validators have and process the same blockchain data locally.
● It's not parallel processing, so it doesn't scale with more validators.

56. Cross-shard Transactions

57. Academic Sharding Research
Year Paper Title Conference
2016 A Secure Sharding Protocol For Open Blockchains CCS 2016
2017
OmniLedger: A Secure, Scale-Out, Decentralized Ledger via Sharding S&P 2018
Chainspace: A Sharded Smart Contracts Platform NDSS 2018
2018
RapidChain: Scaling Blockchain via Full Sharding CCS 2018
Towards Scaling Blockchain Systems via Sharding SIGMOD 2019
2019 Monoxide: Scale Out Blockchain with Asynchronous Consensus Zones NSDI 2019
etc.

58. “Week in Ethereum News”
https://weekinethereumnews.com/week-in-ethereum-news-september-6-2020/

59. “What’s New in Eth2”
https://hackmd.io/@benjaminion/eth2_news/https%3A%2F%2Fhackmd.io%2F%40benjaminion%2Fwnie2_200905

60. Feedback by Other Resercher