The guide to access USB over TCP and UDP network

For the correct operation of USB Network Gate, you’ll need to open some ports:

TCP 5473 – to obtain the detailed list of all shared USB peripherals;
UDP 5474 – to broadcast, for automatic detection of locally shared networked devices;
UDP 5475 – to get an answer to a broadcast query;
TCP 5475 – to manage remote operation;
TCP port that was determined in the server’s sharing options.

UsbConfig.exe and UsbService.exe are using those ports in Windows (or UsbService64.exe in a 64-bit OS).

TCP guarantees a dependable and organized data transfer from server to user and the other way around. UDP is far less thorough and goes on with data transmission regardless of the receiver’s availability.

Reliability

TCP is deemed fail-safe, thanks to packet tracking and resending them if damaged or missing so no data is lost. UDP doesn’t guarantee data delivery, nor does it check for time-outs. It also has no retransmission function.

Data ordering

TCP numbers data packets and sends them one after another in sequence, so they arrive at the receiver in the very same order. If any bit of data arrives out of sequence, TCP will reorder and retransmit that packet. UDP doesn’t bother to sequence data, so packets may arrive at the receiver in any order.

Connection specifics

TCP is heavy-weight. It requires 3 packet exchanges to establish a connection. Also it is very reliable, and carries out congestion control. UDP is lightweight. It’s a transport layer built on top of IP that doesn’t order sent packets or track connections.

Transition method

TCP has a variable-length header (20-80 bytes) and treats data as a byte stream transmitted to segment boundaries. UDP has a fixed 8-byte length header, and sends each packet individually, verifying its checksum upon arrival.

Error checking

UDP only has a basic error detection algorithm. If the checksum of the received packet is incorrect, it just gets discarded with no attempts to request a retransmission. This is why UDP is mostly used for online gaming, live streaming, or other situations when speed is critical while missing a packet or two is not.

TCP has extensive error detection and recovery mechanism, called Positive Acknowledgement with Retransmission (PAR). It immediately resends packet if the recipient reports it to be lost, erroneous, or having a wrong checksum.

TCP and UDP operation

To establish a connection, TCP uses a method called ‘a three-way handshake’. It’s a three-step exchange of SYN and ACK packets between sender and recipient, prior to actual data transfer. Once the data is successfully received, all established virtual links are closed to terminate the connection.

UDP doesn’t need ‘hand-shaking’, for it’s not connection-oriented. Its datagrams may occasionally duplicate, arrive in the wrong order, or just vanish God knows where. While using UDP error detection and fix, if any at all, is carried out at the application level. That allows to unload the network interface level. UDP sustains packet broadcasting (data transmission to everyone on your local network) and multicasting (transmission to everyone who’s currently subscribed).