A solution may solve Block Withholding Attack
3 messages, 3 to 8 Oct 2017. Started by Kevin Pan.
Kevin Pan (2), James Hilliard
Here is a solution may solve Block Withholding Attack. The general idea is came from Aviv Zohar(avivz@cs•huji•ac•il), I made it work for Bitcoin. Anyway, thanks Aviv.
=====================
DIFF_1 = 0x00000000FFFF0000000000000000000000000000000000000000000000000000;
Diff = DIFF_1 / target
this is equal to
Diff = DIFF_1 / (target - 0) or Diff = DIFF_1 / abs(target - 0)
now, we change diff algo to below:
New_Diff = DIFF_1 / abs(target - offset)
Offset is 32 bytes, like uint256 in Bitcoin, range is [0, 2^256), define: offset_hash = DSHA256(offset).
we need to do a little change to the merkle root hash algo, put the offset_hash as a tx hash in the front of tx hashes.
[offset_hash, coinbase_tx_hash, tx01_hash, tx02_hash, … , tx_n_hash]
Actually could put offset_hash in any place in the array of hashes.
network_hash_range = network_hash_end - network_hash_begin
miner_hash_range = miner_hash_end - miner_hash_begin
The offset value MUST between network_hash_begin/end or miner_hash_begin/end.
https://user-images.githubusercontent.com/514951/31133378-e00d9ca2-a891-11e7-8c61-73325f59f6ed.JPG
When mining pool send a job to miners, put the PoW hash range (miner_hash_begin/end) in the job. So if the miners find a hash which value is between [miner_hash_begin, miner_hash_end], means it's SHOULD be a valid share, could submit the share to the pool. If the hash value is between [network_hash_begin, network_hash_end] means find a valid block.
The network_diff is much much high than the miner's diff, means the network_hash_range is much much smaller than miner_hash_range. By now, a typical miner's pool diff is around 16K, network diff is 1123863285132, so miner_hash_range is at least million times bigger than network_hash_range. The miners only know miner_hash_range, it's impossible for cheat miners to find out which share could make a valid block or not.
Problems: 1. it's a hard fork. 2. will make existed asic dsha256 chips useless, but I think it's only a small change to make new asic chips based on existed tech.
There have been some other proposals to deal with this such as https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2012-June/001506.html that may be possible to implement in existing miners.
On Tue, Oct 3, 2017 at 9:52 AM, 潘志彪 via bitcoin-dev <bitcoin-dev@lists•linuxfoundation•org> wrote:
Quoted text (61 lines)
> Here is a solution may solve Block Withholding Attack. The general idea is > came from Aviv Zohar(avivz@cs•huji•ac•il), I made it work for Bitcoin. > Anyway, thanks Aviv. > > ===================== > > DIFF_1 = 0x00000000FFFF0000000000000000000000000000000000000000000000000000; > > Diff = DIFF_1 / target > > this is equal to > > Diff = DIFF_1 / (target - 0) or Diff = DIFF_1 / abs(target - 0) > > now, we change diff algo to below: > > New_Diff = DIFF_1 / abs(target - offset) > > Offset is 32 bytes, like uint256 in Bitcoin, range is [0, 2^256), > define: offset_hash = DSHA256(offset). > > we need to do a little change to the merkle root hash algo, put the > offset_hash as a tx hash in the front of tx hashes. > > [offset_hash, coinbase_tx_hash, tx01_hash, tx02_hash, … , tx_n_hash] > > Actually could put offset_hash in any place in the array of hashes. > > network_hash_range = network_hash_end - network_hash_begin > > miner_hash_range = miner_hash_end - miner_hash_begin > > The offset value MUST between network_hash_begin/end or > miner_hash_begin/end. > > https://user-images.githubusercontent.com/514951/31133378-e00d9ca2-a891-11e7-8c61-73325f59f6ed.JPG > > When mining pool send a job to miners, put the PoW hash range > (miner_hash_begin/end) in the job. So if the miners find a hash which value > is between [miner_hash_begin, miner_hash_end], means it's SHOULD be a > valid share, could submit the share to the pool. If the hash value is > between [network_hash_begin, network_hash_end] means find a valid block. > > The network_diff is much much high than the miner's diff, means the > network_hash_range is much much smaller than miner_hash_range. By now, > a typical miner's pool diff is around 16K, network diff is 1123863285132, > so miner_hash_range is at least million times bigger than > network_hash_range. > The miners only know miner_hash_range, it's impossible for cheat miners > to find out which share could make a valid block or not. > > Problems: > 1. it's a hard fork. > 2. will make existed asic dsha256 chips useless, but I think it's only a > small change to make new asic chips based on existed tech. > > _______________________________________________ > bitcoin-dev mailing list > bitcoin-dev@lists•linuxfoundation•org > https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev >
But I think this one is simpler and better than Luke's.
And now is different like 2012, pools need be more independ today. Pools want to express their opinion or standpoint. Some of can't do that like remove the NYA tag and one the reason is the Block Withholding Attack.
Kevin Pan
On Fri, Oct 6, 2017 at 10:36 PM, James Hilliard <james.hilliard1@gmail•com> wrote:
Quoted text (73 lines)
> There have been some other proposals to deal with this such as > https://lists.linuxfoundation.org/pipermail/bitcoin-dev/ > 2012-June/001506.html > that may be possible to implement in existing miners. > > On Tue, Oct 3, 2017 at 9:52 AM, 潘志彪 via bitcoin-dev > <bitcoin-dev@lists•linuxfoundation•org> wrote: > > Here is a solution may solve Block Withholding Attack. The general idea > is > > came from Aviv Zohar(avivz@cs•huji•ac•il), I made it work for Bitcoin. > > Anyway, thanks Aviv. > > > > ===================== > > > > DIFF_1 = 0x00000000FFFF0000000000000000000000000000000000000000000000 > 000000; > > > > Diff = DIFF_1 / target > > > > this is equal to > > > > Diff = DIFF_1 / (target - 0) or Diff = DIFF_1 / abs(target - 0) > > > > now, we change diff algo to below: > > > > New_Diff = DIFF_1 / abs(target - offset) > > > > Offset is 32 bytes, like uint256 in Bitcoin, range is [0, 2^256), > > define: offset_hash = DSHA256(offset). > > > > we need to do a little change to the merkle root hash algo, put the > > offset_hash as a tx hash in the front of tx hashes. > > > > [offset_hash, coinbase_tx_hash, tx01_hash, tx02_hash, … , tx_n_hash] > > > > Actually could put offset_hash in any place in the array of hashes. > > > > network_hash_range = network_hash_end - network_hash_begin > > > > miner_hash_range = miner_hash_end - miner_hash_begin > > > > The offset value MUST between network_hash_begin/end or > > miner_hash_begin/end. > > > > https://user-images.githubusercontent.com/514951/ > 31133378-e00d9ca2-a891-11e7-8c61-73325f59f6ed.JPG > > > > When mining pool send a job to miners, put the PoW hash range > > (miner_hash_begin/end) in the job. So if the miners find a hash which > value > > is between [miner_hash_begin, miner_hash_end], means it's SHOULD be a > > valid share, could submit the share to the pool. If the hash value is > > between [network_hash_begin, network_hash_end] means find a valid block. > > > > The network_diff is much much high than the miner's diff, means the > > network_hash_range is much much smaller than miner_hash_range. By now, > > a typical miner's pool diff is around 16K, network diff is 1123863285132, > > so miner_hash_range is at least million times bigger than > > network_hash_range. > > The miners only know miner_hash_range, it's impossible for cheat miners > > to find out which share could make a valid block or not. > > > > Problems: > > 1. it's a hard fork. > > 2. will make existed asic dsha256 chips useless, but I think it's only a > > small change to make new asic chips based on existed tech. > > > > _______________________________________________ > > bitcoin-dev mailing list > > bitcoin-dev@lists•linuxfoundation•org > > https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev > > >