Anonymous communication was proposed by Chaum, who put forward an anonymous communication algorithm based on Mix nodes. A Mix node receives messages from multiple senders, mixes and processes these messages, and then transmits them to the receivers, thereby concealing the identity information of the senders and receivers and achieving anonymity.
1. The Basic Framework of Anonymous Communication
1.1 Anonymity Properties
Anonymity properties include unidentifiability and unlinkability. Unidentifiability means that the adversary cannot identify a user’s identity and behavior; unlinkability means that the adversary cannot associate messages, behaviors, and users through observing the system.
1.2 Adversary Capabilities
An adversary is a communication network user, or a set of users, who intends to reduce or eliminate communication anonymity. Anonymous communication systems generally indicate the adversary capabilities they can resist by proposing a threat model (thread mode). Adversary capability falls into three aspects: reachability, attackability, and adaptability.
1.3 Network Types
The network type of an anonymous communication system is determined by the following three factors: path topology, route scheme, and path type.
2. Current Classification of Anonymous Communication:
2.1 Anonymous Communication Systems Based on the Mix Algorithm
The core idea of this class of communication system is to use a single Mix node, or multiple Mix nodes in a cascade, to achieve anonymous communication. A Mix node is a node in the network that provides anonymous communication services to other nodes; it receives data encrypted with its public key, and processes that data by decrypting it, batching it, reordering it, and adding redundant bytes, before transmitting it to the next Mix or to the final receiver. Anonymous communication systems based on the Mix algorithm have the following characteristics:
- Part of the nodes in the anonymous communication network provide anonymous communication services to other nodes;
- The initiator must determine the entire transmission path of the communication before initiating anonymous communication, and that path does not change during transmission;
- The initiator must obtain information about each Mix node along the entire transmission path before initiating anonymous communication, including address and key information;
- Mix nodes process communication information from multiple senders by decrypting, multiplexing, batching, reordering, and adding redundant bytes, giving the system high anonymity, but the communication transmission latency is relatively high, which generally makes it unsuitable for real-time data communication.
Anonymous communication systems based on the Mix algorithm include Babel, Cyberpunk (Type I), Mixmaster (Type II), and Mixminion (Type III).
2.2 Anonymous Communication Systems Based on the Onion Routing Algorithm
Anonymous communication systems based on the Onion Routing algorithm place more emphasis on the real-time nature of data communication, as well as the simplicity, effectiveness, and practicality of the system. Their characteristics are:
- Anonymous communication systems based on the Onion Routing algorithm are built on top of TCP transport, and nodes usually communicate over SSL;
- Anonymous communication systems based on the Onion Routing algorithm use asymmetric key algorithms for encryption when a path is established, and symmetric key algorithms for encryption during data communication, in order to improve data transmission efficiency and reduce latency;
- Anonymous communication systems based on the Onion Routing algorithm use real-time multiplexing and forwarding, and do not perform batching such as reordering communication data or fixing input and output traffic. Anonymous communication systems based on the Onion Routing algorithm include Tor, FreeNet, and others.
2.3 Anonymous Communication Systems Based on Flooding Algorithms
Anonymous communication systems based on flooding algorithms are a recent research hotspot in the field of anonymous communication transport. They mainly achieve anonymous communication through flood-like algorithms such as flooding and epidemic, and they are still at the laboratory research stage, with no mature anonymous communication system actually deployed. Anonymous communication systems based on flooding algorithms generally have the following characteristics:
- Before initiating anonymous transmission, the initiator has no idea of the anonymous transmission path and needs no information about the intermediate nodes of the transmission;
- The path of each anonymous transmission by the initiator is not fixed;
- No intermediate node in the anonymous communication network knows the initiator and receiver of the anonymous communication.
3. Introduction to Tor
Tor is the second-generation anonymous communication system based on the Onion Routing algorithm. It is currently widely deployed across the Internet through volunteer relay nodes, and it is the most successful public anonymous communication service on the Internet. The Tor network has more than 1,000 relay nodes worldwide, most of them located in Germany and the United States, along with millions of users. Torproject is its application project.
Tor is a low-latency anonymous communication service based on circuit switching. Tor’s design introduces mechanisms such as perfect forward secrecy, congestion control, directory service, integrity checking, and configurable exit policies, which solved the various problems in the design of first-generation anonymous communication systems based on the Onion Routing algorithm. Tor has two kinds of entity: the Tor user and the Tor node. A Tor user runs an Onion Proxy (OP) program on the local system; this program is responsible for establishing circuits, receiving the application’s TCP data stream, and transmitting that data stream over an already-established circuit.
The circuit establishment process is as follows.
The OP accesses the directory service and obtains information about the Tor nodes in the network, including IP address, public key, exit policy, bandwidth, and uptime.
The OP randomly selects three Tor nodes as relay nodes: an entry node, a middle node, and an exit node. Among the relay nodes, only the entry node knows the identity of the communication initiator, so how the entry node is chosen is very important for protecting the anonymity of the communication initiator. The middle node knows the identities of the entry node and the exit node in the circuit, but does not know the identities of the initiator and receiver of the anonymous communication. The exit node acts as a gateway, responsible for the application-layer connection between the Tor network and the external Internet, and serves as the relay between the encrypted Tor network traffic and the unencrypted Internet traffic. The exit node knows the identity of the receiver of the anonymous communication. When the OP builds a circuit, the OP negotiates a shared session key with each relay node. Under this design, no single node in the circuit knows the identities of the initiator and receiver of the anonymous communication, so communication anonymity is achieved.
The communication message transmission process is as follows.
Once the circuit is established, the OP can begin transmitting application data. The OP receives application data through the SOCKS protocol, and then selects the most recently established circuit for transmission. During communication, the OP splits application messages into 512 B cells, and each cell is in turn encrypted with the session keys shared between the OP and the relay nodes, in the order: exit node, middle node, and entry node. As data travels through the circuit, the relay nodes use the session keys to decrypt it, and after decryption pass it on to the next relay node. The exit node restores the message to plaintext and transmits it to the receiver.
Tor is suited to low-latency anonymous communication that has both an anonymity requirement for data transmission and a real-time requirement for data transmission, such as web access and instant messaging.
