Wireshark and Packet Capture

Networking · Last updated · About 5 min read

Overview

Packet capture means recording the actual network traffic passing a point in the network, packet by packet, for inspection. For radio-system support it is the ground truth: configuration screens and status LEDs can lie, but a capture shows exactly what a gateway, console, or server sent and received. Wireshark is the standard free, graphical capture-and-analysis tool (Windows, macOS, Linux); tcpdump is its command-line counterpart on Linux and embedded devices. Both read and write the same capture file formats (pcap/pcapng), so a capture taken with one can be analyzed with the other.

Capturing at the Right Point

A capture only shows traffic that reaches the capture interface. Modern switched networks deliver each device only its own traffic, so where you capture matters more than any other decision:

  • On the PC running the client/console/server: simplest — install Wireshark on the machine at one end of the conversation. You see everything that host sends and receives, which covers most RoIP (Radio over IP) client-versus-gateway questions.
  • SPAN/mirror port: configure a managed switch to copy ("mirror") traffic from the port serving the device of interest (for example, an embedded gateway you can't install software on) to a port where your capture PC listens. Under heavy load a SPAN port can silently drop copies, so treat marginal loss measurements with care.
  • Inline tap: a hardware device inserted into the cable run that passes traffic through while copying it to a monitor port. The most faithful method, at the cost of briefly interrupting the link to install it.

Capture as close as possible to the device in question; if the question is "is the WAN losing packets," capture at both ends simultaneously and compare.

Essential Display Filters

Wireshark distinguishes capture filters (limit what is recorded) from display filters (limit what is shown after the fact). Prefer capturing broadly and filtering on display. The handful of display filters that cover most support work:

  • ip.addr == 192.0.2.10 — all traffic to or from one host
  • ip.src == 192.0.2.10 / ip.dst == 192.0.2.10 — one direction only
  • udp.port == 5060 / tcp.port == 443 — traffic on a given port
  • udp or tcp — one transport only (see Networking UDP, Networking TCP)
  • rtp — packets Wireshark has recognized as Real-time Transport Protocol voice
  • Combine with &&, ||, ! — e.g. ip.addr == 192.0.2.10 && udp

Recognizing Voice Streams

Live voice has an unmistakable capture signature: a steady, evenly spaced stream of small UDP packets (commonly around 50 packets per second per direction, roughly 60–220 bytes each depending on codec) flowing between two hosts for the duration of a transmission.

Wireshark's RTP dissector unlocks analysis: if the stream isn't auto-detected, right-click a packet and use "Decode As…" to mark the UDP port as RTP. Then the Telephony menu's RTP stream analysis lists each stream with packet counts, loss, and jitter statistics, and can play back audio for common codecs such as G.711. Note that some RoIP products use proprietary framing that only resembles RTP — the steady-small-UDP-packet pattern still identifies the voice flow even when full dissection isn't possible.

Diagnosing Common Voice Problems

  • One-way audio: filter to the two endpoints and look at direction. Healthy audio shows two opposing streams; one-way audio shows packets flowing in only one direction past your capture point. Capture at the receiving end to prove whether the missing stream ever arrives — if it left the sender but never arrived, a firewall, NAT (Network Address Translation), or routing device in between ate it.
  • Packet loss: RTP sequence numbers increment by one per packet; gaps are drops. Wireshark's RTP analysis reports loss percentage per stream. Sustained loss above roughly one percent is audible.
  • Jitter: the same analysis reports interarrival jitter. Highly uneven packet spacing visible in the time column tells the same story. See Latency, Jitter, and Packet Loss in Voice Networks.
  • No traffic at all: the application never sent, sent to the wrong address, or a host firewall blocked it before it reached the wire — check the sending host's own capture and its firewall rules.

Capturing on Windows

Wireshark on Windows requires the Npcap packet-capture driver, installed with Wireshark by default. Practical notes:

  • Capturing usually requires administrator rights (granted to Npcap at install time).
  • Choose the correct interface — PCs with VPN adapters, virtualization adapters, and Wi-Fi plus Ethernet have many; the live traffic-rate sparkline next to each interface helps identify the busy one.
  • A VPN complicates captures: on the physical adapter, tunneled traffic appears encrypted; capture on the VPN's virtual adapter to see the cleartext inside. See Networking VPNs.
  • If capturing on a VLAN (Virtual LAN) trunk or needing VLAN tags preserved, be aware some NIC drivers strip tags before capture — see Networking VLAN.

Etiquette and Handling of Capture Files

A capture records everything crossing the interface: credentials in cleartext protocols, emails, and playable voice audio. Treat capture files as sensitive material — share them only with people authorized to see the traffic, store them accordingly, and delete them when the case closes. When sending a capture to a vendor, filter it down to the relevant hosts/ports first, and say what time window and symptom the capture covers.

References

This article is part of the CTI Products Knowledge Base. Questions about RadioPro or MCN products? Call +1 (513) 595-5900 or email info@ctiproducts.com.

Informational only. This article is general background and is not engineering, safety, legal or regulatory advice. Verify requirements with the equipment manufacturer and the applicable authority before relying on it. Product names and trademarks belong to their respective owners.