Network Guideline

In this chapter we want to give you an short overview of important network parameters and the recommended Network architectures.

Basic Network Parameters

Bandwidth

Bandwidth refers to the maximum amount of data that can be transmitted over a network connection within a specified period. It is often measured in bits per second (bps). Higher bandwidth allows for faster data transfer.

Packet Size

Packet size refers to the amount of data transmitted in a single data packet. Data is divided into packets for transmission, and the size of a packet can vary depending on the protocol. Larger packets can be more efficient for transmitting large amounts of data, while smaller packets may be advantageous in high-latency situations.

Tip

Bandwidth influences the camera’s transmission quality. The use of Jumbo Frames can enhance efficiency compared to standard Ethernet packets but requires corresponding network configuration.

Overhead

Overhead refers to additional data or resources required for the management and control of the network but do not directly contribute to the payload. This may include protocol information, error corrections, or other control information. Higher overhead can reduce effective bandwidth.

Throughput

Throughput is the actual amount of data that can be successfully transmitted over a network connection. It takes into account the impact of overhead, latency, and other factors. Throughput is often measured in the same units as bandwidth, such as bits per second. It’s important to note that throughput is typically lower than the nominal bandwidth due to network delays, collisions, overhead, and other factors.

Overview Network Architectures

Direct Camera – NIC connection (recommended)

In this case the camera is connected directly to a network card (NIC) using a GigE Vsion cable. This configuration is the most efficient connection when the connection is exclusive to the dedicated device, ensuring that the entire bandwidth is allocated to the camera. As long as the network card is appropriately set to full-duplex mode, Ethernet data operates without collisions.

However, a collision can still occur if the cameras compete for the host resources. In this case, the host may not be able to process the incoming data if all cameras are transmitting at the same time, which can lead to packet loss.

To get an overview of the recommended network adapter settings, see chapter Network Settings.

Tip

This is the network architecture we recommend for our AT sensors and cameras.

Multiple cameras over one or more switches

Utilizing multiple cameras through one or more dedicated switches involves connecting the cameras, or even a single camera, not directly to the host but through a switch. However, the entire network remains exclusively dedicated to the purpose of image acquisition. In more intricate setups, various switches and multiple hosts can be employed. These switches facilitate the connection of multiple cameras to the host PC and may extend the distance between the camera and the host.

In this scenario, cameras share common network segment(s) and the switch itself. They not only contend for host resources but also for Ethernet bandwidth and switch resources. The introduction of additional network components increases system latency and jitter. Careful consideration is necessary when selecting a switch model to ensure it does not impose undesirable constraints on the overall system configuration.

Shared Network for General Purposes

In this scenario, the image processing camera shares the network with the Internet, the office or other types of traffic. Generally, this configuration is not ideal for machine vision applications and not recommended, as the presence of various network traffic can significantly impact image acquisition performance. It is advisable to consider this configuration only in low-performance applications where a dedicated network segment is either unsuitable or not feasible.

Tip

Choosing network cards and switches must align with the camera’s requirements. Key configuration settings include Jumbo Frames and Flow Control for optimal performance.

Why it is recommended to not use a network switch?

After the camera has captured an image, the data is transmitted to the network. The packets pass trough the network towards the host PC. While passing through a switch, temporary blockages may occur if the switch is, for example, occupied processing packets from other cameras. If the switch is overloaded with incoming data, packet losses may occur.

Upon reaching the host’s network card, the packets are accumulated, signaling the presence of new data through interrupts. The data is then processed by either a standard operating system network stack or a dedicated (filter) driver, and it is finally delivered to the target buffer. If the host is under a heavy load and cannot manage the incoming data rate, there is a possibility of dropping packets once again.