Extract: All of the data networks available today have serious flaws, vulnerabilities or inconveniences. We propose a solution that could, in a significant (if not all) of the cases, replace these methods by eliminating their flaws. The network is similar to the torrent protocol in many ways, but its users do not have to contribute to its operation at all. To make it worthwhile for some people to continue to maintain thenetwork, the network issues a cryptocurrency. The nodes that forward the data packets are rewarded for the amount of data they forward, and their work is verified by other nodes through a consensus process called proof-of-success. Data packet headers and transactions through the network would be stored in a blockchain and a procedure would be used to maintain the pseudo-randomness of the network selectionmechanism, thus minimizing the number of possible abuses.
Today, there are 3 common methods for sharing data packets over the Internet. The simplest is the peer-to-peer model, where two computers can communicate directly without a third party. This is undoubtedly the most secure solution, but unfortunately it is severely limited by the need for all parties to be online from start to finish of the data transfer, and by the difficulty of implementing it with a large number of users. This problem is solved by the server-client model, which has enabled giants such as Amazon, Google and Microsoft to penetrate every part of our lives. While centralised server-based solutions are mostly subscription-based, the last decade has seen the proliferation of targeted advertising, which means that Facebook and Google now know more about us than our friends. Cyber attacks in recent years have shown that our data is far from in good hands, partly due to human error and government regulations that require service providers to give intelligence and law enforcement agencies access to data. In order to safely use the all-encompassing internet of the future, it is necessary to have a transparent network independent of any centralized third party, which is also capable of performing the tasks of today's systems and does not require contributions from the user side. In this paper, we propose a solution that sends data packets through a number of pseudo-randomly chosen nodes in exchange for a coin issued by a network, and the computers that oppose their work justify their position by a certain amount of committed coins. The system can remain secure as long as the benign nodes have more coins than the malicious ones.
As shown in the figure below, the network is composed of 4 types of actors: endpoints (i.e. users), relay servers, validators and leader nodes. The soul of the network is the group of the leader nodes, through which the connection data is transmitted to the nodes. The relay servers are responsible for forwarding the data packets, while the validators, which operate in groups, communicate to the leader nodes the success of the data packets in reaching their destinations.
Binary data packets of arbitrary size are delivered to the receiver in multiple chunks. This has two advantages. Firstly, nodes cannot be forced to delete the data that passes through them, so if the cryptographic algorithm used for encryption were to be cracked or a malicious party were to gain a significant quantum advantage - it would take 8 hours to crack a 2048-bit RSA key with 20 million quantum bits - the contents of all messages ever sent could potentially be leaked. In addition, breaking messages into multiple chunks helps distribute the load across nodes, thereby reducing network latency. Since encryption with RSA is severely limited both in time and in maximum size, we propose a solution, already used in several places, referred to as hybrid encryption, where the data itself is encrypted using a symmetric method (e.g., AES) and then RSA public-private key-based encryption is applied on the significantly shorter key. It is important that the fragments in a stream encryption method are not contiguous bits, as meaningful details can be extracted from them if the key is known. Instead, it makes sense to extract bits based on a pattern known to both parties, thus avoiding this problem. A further problem is that if for some reason any chunk of the message fails to reach its destination, decoding will be impossible, so it is important to allow for redundancy by routing individual chunks to multiple nodes.
To reward relay servers fairly, we must assume that relay servers are malicious, thus they want to minimize the amount of data transferred and maximize the reward. In order to check whether the data packet actually reaches the receiver, we assign a set of n servers to each packet, who then decide on a consensus basis whether the delivery of the data packet was successful. Only the IP addresses of the validators and the receiver are transmitted to the relay server, so the relay server does not know which one is the validator and which one is the receiver. However, it would still be sufficient to forward the data packet to n/2+1 IP addresses, which in reallity does not ensure that the receiver will receive the message. These odds can be overridden by randomly allocating n/2 more votes among the validators, and since the relay server is not aware which validators have more votes, it is forced to forward the data packet to all IP addresses. In addition, we can further reduce the susceptibility to fraud by crediting the rewards collected by the relay servers only after a certain amount of data has been transmitted and if they try to cheat the system, they lose the locked amount.
The process, first described by Satoshi Nakamoto in his Bitcoin white paper, is nowadays mainly used for cryptocurrencies. The idea is to sort the data into blocks (e.g. incoming transactions in the last 3 minutes), apply a hash function to this block and hash the next block with the result. The result is that if we change even one bit in any element of the chain, we get a completely different result at the end of the chain. Since today's average CPU can perform thousands of hashes, it is easy to iterate through the chain and check its correctness. In the case of our network, we store not only transactions, but also data packet headers, which include the sender, receiver, number of cracks, and packet length. Thanks to blockchain technology, this data can be maintained securely without being altered, thus guaranteeing the correctness of the reward.
The system can only stay reliable without the involvement of third parties if an easily definable, traceable and universally definable procedure is used to select each node. The basis of the procedure is a hash function (in our case, we propose SHA-256). The goal of the procedure is to select a pseudorandom but unambiguous node from a certain data set, i.e. a node with the same parameters will always have the same result. For the selection of leader nodes, only the hash of the previous block is input, but for the selection of relay servers and validators, the header of the given data packet is included too, since their selection is for a single data packet, not for the duration of a whole block. From the hash obtained from this data packet, we crack off a fraction of the size of the largest integer that can be formed, which is at least a few orders of magnitude larger than the expected total number of nodes in the network. Let this size be 32 bits for the time being. For each node reporting for that position, we allocate an index in sequence from zero, and take the modulus of the 32-bit number subtracted from the beginning of the hash and the number of nodes reporting for that position, which will be the index of the node thus selected. The method can be used to select multiple nodes that are pseudorandomly but uniquely selected, for example an entire validator group, by appending indices from zero to the original data packet before hashing. The Python code of the procedure looks like this (names: n1, n2, s1, s2 =
number of nodes needed, total number of nodes, last block hash, message header):
def chooseNodes(n1, n2, s1, s2 = ''):
chooseNodeIndexes = [];
for i in range(n1):
choosenNodeIndexes.append(int(SHA256(s1+s2+i)[0:32]) % n2);
return choosenNodeIndexes;Relay servers cannot harm the network, but the leader nodes and the validators have a direct impact on the blockchain, so trust is essential. For many cryptocurrencies, trust is provided by a process called Proof-of-Work, i.e. a random number is placed in the block, which causes the hash of the block to start with a given number of 0 bits. Whoever finds this random number first gets the reward and iscredited by the network. This means that the individual with the greatest computational power dominates the network, which means that in the near future, individuals with quantum dominance could easily take control of the network. It is also extremely energy and time intensive, making it difficult to sustain in the long term. In our case, the network requires a certain amount of coins to be deposited as a token of trust. Since validators and leader nodes work in groups, the consensus result of the groups will be permanent on the blockchain, while those who voted against the majority will lose their staked coins, and those who voted with the majority will receive interest on the coins. The only modification to the procedure described above is that an individual can be added to the lists more than once, so the more coins someone has, the more likely they are to be selected and the more rewards they can earn. This means that the person with more than half of the coins, which is virtually impossible to obtain, can gain complete control of the network.