Skip to content

Computer Networking: OSI, TCP/IP, IPv4/IPv6 and Subnetting

Networking from signalling to subnetting for digital forensic examiners in India: OSI vs TCP/IP, IPv4 classes and CIDR, IPv6 SLAAC, MAC addressing, switching, routing, Wi-Fi 7, and which header fields survive each layer of a packet capture under CERT-In's 180-day log retention rule.

Last updated:

Share

Computer networking for forensic examiners rests on two complementary models: the seven-layer OSI reference model, which defines how network functions are partitioned and labelled in tools such as Wireshark, and the four-layer TCP/IP stack, which is the protocol suite actually deployed on the internet. Every frame a packet-capture examiner opens contains nested headers from both models: a destination MAC at Layer 2, a source IP at Layer 3, a TCP port at Layer 4, and a TLS Server Name Indication at Layer 7, each with its own trust basis and its own survival rules across NAT, routing, and encryption boundaries. Under CERT-In Direction 2022, Indian service providers must retain these layer-by-layer log artefacts for 180 rolling days, and correctly framing a log request requires knowing which header field lives at which layer.

Every PCAP an investigator opens at a state cyber cell carries seven layers of nested headers. A defence counsel under BSA Section 63 may ask why a destination MAC, a source IP, a TCP sequence number and a TLS Server Name Indication can all be trusted as belonging to the same conversation. The answer lies in the OSI model, the TCP/IP stack that runs the internet, and the addressing rules that determine who can communicate with whom. Errors at any layer undermine attribution throughout the rest of network forensics.

Key takeaways

  • The OSI model partitions network function into seven layers, each with its own header, giving an examiner seven distinct points at which an actor can be identified in a packet capture.
  • CERT-In Direction 2022 mandates 180 days of log retention for service providers, VPNs, and data centres, and those logs are the primary records a digital forensic examiner interprets when a case arrives.
  • A destination MAC, a source IP, a TCP sequence number, and a TLS Server Name Indication can all appear in the same captured frame but each belongs to a different protocol layer with its own trust basis.
  • IPv6 transition changes addressing assumptions forensic examiners carry from IPv4 cases, requiring familiarity with both stacks to correctly attribute traffic in mixed-protocol environments.
  • Bandwidth and throughput differ: a 1 Gbps link's real throughput is lower due to framing overhead, retransmissions, and TCP slow-start, a distinction that matters when estimating data exfiltration volumes from flow records.

This topic is the digital forensics networking foundation, written for an examiner who already knows what a hard disk is and now needs to read traffic. It walks signalling through subnetting through routing, with the Indian operational frame layered on top: CERT-In Direction 2022 mandates 180 days of mandatory log retention for service providers, VPNs and data centres, and those logs are what the digital forensic examiner has to interpret when the case lands. The Wi-Fi standards, MAC address handling, and IPv6 transitions covered here feed directly into the wireless attack topic and the live packet capture topic.

By the end of this topic you will be able to:

  • Identify the seven OSI layers and the corresponding TCP/IP layer, and name at least one protocol and one forensic artefact recoverable from a packet capture at each layer.
  • Explain how encapsulation and decapsulation work across a multi-hop path, including which header fields (MAC, IP, port, TLS SNI) survive NAT and routing and which are rewritten.
  • Calculate host counts and usable address ranges from CIDR notation, and distinguish RFC 1918 private ranges, APIPA, and CGNAT from public addresses.
  • Describe IPv6 SLAAC address formation using EUI-64 vs. RFC 7217/8981 privacy identifiers, and state what each form does or does not reveal about the originating device.
  • Frame a precise CERT-In log request that specifies the correct fields (IPv4 NAT mapping with source port, IPv6 prefix delegation, NDP cache snapshot) for a dual-stack Indian ISP environment.
Key terms
OSI reference model
Seven-layer ISO/IEC 7498-1 model describing how network functions are partitioned: Physical, Data Link, Network, Transport, Session, Presentation, Application. A teaching model, not a deployed stack.
TCP/IP stack
Four-layer model that actually runs the internet: Link, Internet, Transport, Application. Defined by the IETF and mapped onto the OSI layers for analytical use.
CIDR (Classless Inter-Domain Routing)
Notation `a.b.c.d/n` where the trailing slash gives the prefix length. Replaced classful addressing in 1993 (RFC 1518/1519) and enabled supernetting and route aggregation.
RFC 1918 private space
Three IPv4 ranges reserved for private use: 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16. Not routable on the public internet; require NAT at the edge.
MAC address (OUI)
48-bit hardware address. First 24 bits are the Organisationally Unique Identifier assigned by IEEE to a vendor; the lower 24 are the vendor's own serial.
SLAAC
Stateless Address Autoconfiguration. IPv6 host generates its own address from the router-advertised /64 prefix plus a 64-bit interface identifier (EUI-64 or RFC 7217 stable random).

Signalling, bandwidth and the wires under the model

Every network conversation begins as a physical signal. Two coding choices govern that signal. Digital signalling represents data as discrete voltage levels (a Manchester-encoded Ethernet frame, NRZ-L on a serial line); analog signalling represents data as a continuously varying waveform (the legacy POTS line that still feeds many Indian rural exchanges, the DOCSIS RF carrier on a cable modem). Modern LANs use digital signalling because it tolerates noise far better and allows the receiver to re-time itself from the bit clock.

The second axis is baseband versus broadband. Baseband dedicates the entire medium to a single signal at a time, as 10BASE-T, 100BASE-TX, 1000BASE-T and 10GBASE-T do ("BASE" is short for baseband). Broadband frequency-divides the same medium so that multiple signals run in parallel on different sub-carriers, as DOCSIS cable, ADSL, and OFDMA-based Wi-Fi 6 do. A broadband medium produces independent per-channel error and timing data that an attacker can use to fingerprint a device.

Latency, the round-trip time between two endpoints, is the second performance metric the examiner reads off a capture. On a healthy domestic Jio Fiber link in Mumbai an ICMP RTT to an AWS Mumbai region endpoint sits at around 4 to 8 ms; on a 4G LTE link in a Bengaluru suburb the same ping is 35 to 60 ms; on a satellite link it climbs to 600 ms or more. Jitter is the standard deviation of those RTTs and is what makes a real-time voice call collapse before bandwidth does. The bps unit is bits per second (lowercase b); Bps is bytes per second (uppercase B). An ISP advertises in bps; a file transfer reports in Bps. The factor of eight between them turns up in court transcripts when a defence expert tries to argue an upload was impossibly fast.

The OSI 7-layer model and what forensics gets at each layer

The OSI model partitions network function into seven layers. It is a reference framework, not a deployed protocol; the internet runs on the four-layer TCP/IP stack. The seven-layer breakdown is the labelling scheme used by Wireshark's dissector pane and the standard networking textbook.

#LayerFunctionExample protocolsForensic artefact
7ApplicationUser-facing protocol logicHTTP, HTTPS, FTP, SMTP, DNS, SSHURLs, mail headers, TLS SNI, DNS QNAMEs
6PresentationEncoding, encryption, compressionTLS, ASN.1, JPEG, gzipCipher suite, certificate chain, JA3 hash
5SessionDialogue control between hostsNetBIOS, RPC, SOCKS, QUIC initialSession IDs, RPC method names
4TransportEnd-to-end delivery, ports, flow controlTCP, UDP, SCTP, QUICSource/destination ports, TCP flags, sequence numbers
3NetworkLogical addressing and routingIPv4, IPv6, ICMP, IPsec, OSPF, BGPSource/destination IP, TTL, fragment offset
2Data LinkFrame-by-frame delivery on the local segmentEthernet, 802.11, PPP, ARP, VLAN 802.1QSource/destination MAC, VLAN tag, OUI vendor
1PhysicalBits on the wire or in the air10BASE-T, 1000BASE-T, 802.11ax PHYSignal strength, channel, modulation

Encapsulation is the operation that wraps a payload from layer N inside a layer N minus one header on the way down the stack, and decapsulation strips those wrappers on the way back up. An HTTPS request to nfsu.ac.in on a Wi-Fi 6 laptop in a Gandhinagar reading room becomes, in order: an HTTP method, wrapped in a TLS record, wrapped in a TCP segment with destination port 443, wrapped in an IPv6 packet with the laptop's link-local source and the resolver's destination, wrapped in an 802.11 data frame addressed to the access point's BSSID, wrapped in radio symbols on a 6 GHz channel. The PCAP records every wrapper. The examiner reads them in reverse.

The forensic implication per layer is the heart of network forensics. A MAC address survives only inside the Layer 2 frame; the moment a router forwards the packet, the source and destination MACs are rewritten to those of the next hop. An IP address survives end-to-end only if no NAT sits in between; in any home router or mobile carrier path, the source IP changes at the edge. A port number survives end-to-end through NAT (though the source port is often rewritten under PAT). A Server Name Indication string inside the TLS ClientHello survives end-to-end because TLS is end-to-end. Knowing what survives where is the difference between attributing an attack to a device and attributing it to an autonomous system.

TCP/IP four-layer model and encapsulation in practice

The internet runs TCP/IP, a four-layer stack defined by the IETF before OSI was finalised and the only one that ships in every operating system. The TCP/IP layers map onto OSI for analysis but combine functions that the OSI model separates.

  1. Link layer (OSI 1 + 2)
    The physical medium plus framing. Ethernet, 802.11, PPP. Source and destination MAC live here. ARP runs here on IPv4; NDP runs here on IPv6.
  2. Internet layer (OSI 3)
    Logical addressing and routing. IPv4 or IPv6 header. ICMP, IGMP and IPsec sit here. Routing protocols (OSPF, BGP, RIP) deliver the routing tables that this layer consults.
  3. Transport layer (OSI 4)
    End-to-end delivery. TCP for reliable byte streams with sequence numbers, acknowledgements and a three-way handshake. UDP for connectionless datagrams. QUIC as the modern UDP-based replacement for TCP+TLS, now carrying the majority of HTTPS traffic to Google and Cloudflare.
  4. Application layer (OSI 5 + 6 + 7)
    Everything above transport. HTTP, HTTPS, DNS, FTP, SMTP, IMAP, SSH, BGP, MQTT, gRPC. TLS is sometimes placed here, sometimes between transport and application; in TCP/IP it is treated as application-layer.

The three-way handshake is the diagnostic backbone of TCP forensics. The client sends a SYN with an initial sequence number X; the server replies SYN-ACK with its own initial sequence number Y and acknowledgement X+1; the client replies ACK with Y+1, and the connection is open. A capture that shows a SYN with no SYN-ACK back is a port that is closed or filtered. A SYN-ACK with the wrong sequence number is a sign of in-path tampering. A connection that closes with a RST instead of a FIN is either an application crash or an intentional reset.

A CERT-In incident report under the 2022 Direction has to identify the affected IP address, the source port and the destination port for every observed connection, and the reporting window is six hours from detection. CFSL Hyderabad's cyber lab and the state cyber cells routinely receive PCAPs where the handshake is the only artefact that survives end-to-end NAT, because the IP layer is rewritten at the carrier edge while the port and sequence numbers are not.

Snapshot-based VM forensics workflow: six sequential steps from live VM through artefact extraction, with the volatility-vs-s
OSI 7-layer stack vs TCP/IP 4-layer stack side by side with example protocols at each layer and the forensic artefact a PCAP examiner recovers. Layers highlighted in blue are directly visible in a standard Wireshark capture; the Physical layer is only visible in radio captures (802.11 monitor mode).

The well-known port range 0 to 1023 is reserved by IANA: 22 for SSH, 25 for SMTP, 53 for DNS, 80 for HTTP, 110 for POP3, 143 for IMAP, 443 for HTTPS, 465 for SMTPS, 587 for submission, 993 for IMAPS. The registered range 1024 to 49151 is where most application servers run. The dynamic range 49152 to 65535 is where clients pick their ephemeral source ports. A forensic capture that shows a connection on port 443 with traffic that does not begin with a TLS ClientHello is one of the strongest indicators of protocol smuggling, including a Cobalt Strike beacon or an SSH-over-port-443 tunnel.

IPv4 addressing, CIDR and the private/public split

IPv4 is a 32-bit address written as four dotted-decimal octets, providing a theoretical 4.29 billion addresses. The original 1981 design partitioned that space into classes by the high-order bits. Class A (0.0.0.0 to 127.255.255.255) allocated a /8 to organisations such as the US Department of Defense, Apple and MIT; Class B (128.0.0.0 to 191.255.255.255) allocated a /16; Class C (192.0.0.0 to 223.255.255.255) allocated a /24; Class D (224.0.0.0 to 239.255.255.255) is multicast; Class E (240.0.0.0 to 255.255.255.255) is reserved. The classful scheme allocated far more addresses than most organisations required, wasting large portions of the available space.

CIDR replaced classes in 1993. The CIDR notation 203.0.113.0/24 means the first 24 bits are the network prefix and the remaining 8 bits are the host portion, yielding 256 addresses of which 254 are usable (the first is the network address, the last is the broadcast). The mask 255.255.255.0 is the dotted-decimal equivalent of /24. CIDR allowed prefix lengths between the old class boundaries, so a /22 (1022 hosts) or a /28 (14 hosts) became routine, and address allocation tightened to actual need.

PrefixMaskHostsTypical use
/8255.0.0.016,777,214Legacy Class A (10.0.0.0/8 private)
/16255.255.0.065,534Large campus, RFC 1918 172.16/12 chunks
/22255.255.252.01,022ISP downstream block
/24255.255.255.0254Small office LAN, RFC 1918 192.168.x.0/24
/29255.255.255.2486Point-to-point with spare addresses
/30255.255.255.2522Point-to-point link
/31255.255.255.2542RFC 3021 point-to-point (no broadcast)
/32255.255.255.2551Single host route, loopback advertisement

RFC 1918 reserves three ranges for private use: 10.0.0.0/8 (one /8, the largest), 172.16.0.0/12 (sixteen /16s), 192.168.0.0/16 (256 /24s). These are not routed on the public internet; a router that sees a packet with a 10.0.0.5 destination on its WAN interface drops it. APIPA (169.254.0.0/16) is the link-local block a Windows host self-assigns when DHCP fails. The loopback range 127.0.0.0/8 traditionally carries 127.0.0.1, though the whole /8 is reserved. Indian residential ISPs run carrier-grade NAT on the public IPv4 side and a private /24 inside the home; a subpoena under BNSS Section 94 for "the IP" without the source port and timestamp typically returns thousands of subscribers because they all share the same public address through CGNAT.

VLSM, Variable-Length Subnet Masking, lets a single organisation use different prefix lengths in different parts of its network: a /24 for the office LAN, /30s for the WAN links between routers, a /22 for the campus Wi-Fi. FLSM, Fixed-Length Subnet Masking, uses the same prefix length everywhere and wastes address space. Supernetting, the inverse of subnetting, aggregates contiguous prefixes into a single shorter prefix: two /24s in 203.0.113.0/24 and 203.0.114.0/24 advertised together become 203.0.113.0/23. CIDR's primary operational purpose was route aggregation through supernetting, so that BGP routers carry hundreds of thousands of routes rather than tens of millions.

IPv6, SLAAC, EUI-64 and the dual-stack transition

IPv6 is 128 bits wide, written as eight groups of four hex digits separated by colons. A double colon :: once per address compresses consecutive zero groups. The address 2001:0db8:0000:0000:0000:0000:0000:0001 becomes 2001:db8::1. The address space is 3.4 times 10 to the 38th, enough that exhaustion is no longer the operational concern.

PrefixScopePurposeForensic note
2000::/3Global unicastPublic, routable internet addressEquivalent to a public IPv4 address; logged at carrier
fe80::/10Link-localAuto-configured per interface; not routedAlways present on every IPv6 interface; revealed by NDP
fc00::/7 (fd00::/8 in practice)Unique localPrivate organisation use; not routable globallyIPv6 equivalent of RFC 1918
ff00::/8MulticastOne-to-many deliveryReplaces IPv4 broadcast; e.g. ff02::1 is all-nodes link-local
::1/128LoopbackThe single hostReplaces 127.0.0.1
64:ff9b::/96IPv4-translatedNAT64 prefix per RFC 6052Visible in dual-stack carrier networks

SLAAC, Stateless Address Autoconfiguration (RFC 4862), is how an IPv6 host gets an address without a DHCP server. The router multicasts a Router Advertisement carrying a /64 prefix. The host appends a 64-bit interface identifier and forms its global address. The interface identifier was historically EUI-64, derived from the 48-bit MAC by inserting fffe in the middle and flipping the universal/local bit; the EUI-64 form leaked the device's MAC into every packet, which is a privacy disaster. RFC 7217 specified stable opaque interface identifiers derived from a hash of the prefix, an interface name and a secret, and RFC 8981 specified temporary privacy addresses that rotate every 24 hours. Modern Windows, macOS, iOS and Android all use one of these privacy schemes by default.

The forensic consequence is sharp. An older IPv6 capture (pre-2014) typically shows an EUI-64 interface identifier from which the examiner reads off the device MAC by stripping the inserted fffe and flipping the U/L bit. A modern capture shows an opaque RFC 7217 identifier that reveals nothing about the hardware. To attribute a modern IPv6 host the examiner needs DHCPv6, NDP cache snapshots from the local router or the carrier's IPAM logs.

MAC addressing, switching, routing and the network hardware zoo

A MAC address is 48 bits, usually written as six pairs of hex digits separated by colons or hyphens (a4:83:e7:1c:9d:42). The first three octets (24 bits) are the OUI, the Organisationally Unique Identifier the IEEE assigns to a vendor: 00:1a:11 is Google, a4:83:e7 is Apple, f0:18:98 is Lenovo. Wireshark's OUI lookup, drawn from the IEEE OUI registry, resolves the prefix to a vendor in the packet pane. The second three octets are the vendor's serial. Two bits in the first octet matter: the I/G bit distinguishes unicast (0) from multicast (1); the U/L bit distinguishes universally administered (0, IEEE-assigned) from locally administered (1, software-set). Privacy MAC randomisation in modern phones flips the U/L bit and generates a fresh random MAC per SSID, which is why a captured probe request from a 2023+ iPhone usually shows a locally administered MAC rather than the device's burned-in OUI.

Network hardware splits along OSI layers:

  • Hub (Layer 1): a multi-port repeater that broadcasts every signal to every port. Deprecated since 2000; effectively dead in 2026. Forensic relevance: a true hub creates a passive sniffing surface, which is why packet captures in surveillance setups used to be done at hubs.
  • Bridge (Layer 2): a two-port device that learns MAC addresses and forwards selectively. Replaced functionally by switches.
  • Switch (Layer 2): a multi-port bridge. Maintains a MAC address table, forwards frames only to the port hosting the destination MAC. A managed switch supports VLAN tagging (802.1Q, a 4-byte tag inserted after the source MAC), port mirroring (SPAN/RSPAN, used by SOCs to feed packet brokers), and STP (Spanning Tree, 802.1D, breaks loops by electing a root bridge).
  • Router (Layer 3): inspects the destination IP, consults its routing table, forwards on the egress interface. Rewrites the source and destination MAC at every hop; the IP is preserved.
  • Firewall (multi-layer): operates from Layer 3 up to Layer 7. A stateful firewall tracks TCP connection state; a next-generation firewall does deep packet inspection at Layer 7. See network security: firewalls, IDS, IPsec, SSL/TLS, VPN, PKI, SIEM for how these sit in defence architecture.
  • Access point (Layer 2): the wired-to-wireless bridge. The BSSID is its MAC address. Forensic captures from an enterprise WLC carry per-AP, per-radio metadata.
  • Load balancer (Layer 4 or 7): distributes connections across backend servers. At Layer 4 it sees TCP/UDP ports; at Layer 7 it terminates TLS and reads HTTP headers.

Switching methods differ in how soon a switch starts forwarding a frame. Store-and-forward reads the full frame, validates the FCS, then forwards; high latency, no bad-frame propagation. Cut-through forwards as soon as the destination MAC is read at byte 14; low latency, propagates errored frames. Fragment-free reads the first 64 bytes (one Ethernet collision window) before forwarding; a middle ground used in some Cisco Catalyst products. Most enterprise switches in 2026 run store-and-forward by default.

Routing protocols split by algorithm. RIP (distance-vector, hop count, 15-hop max, RFC 2453) is legacy. EIGRP (Cisco proprietary distance-vector, announced as open in 2013 and documented as RFC 7868 in May 2016) is still in use in mid-size enterprises. OSPF (link-state, Dijkstra, RFC 2328 for v2 / RFC 5340 for v3) is the dominant interior gateway protocol. IS-IS (link-state, ISO 10589) runs the cores of large service providers including Reliance Jio. BGP (path-vector, RFC 4271) is the only exterior gateway protocol; every internet route is carried by BGP. A default route, written as 0.0.0.0/0 on IPv4 or ::/0 on IPv6, is the catch-all the router uses when no more specific match exists; in a home network the default route points at the ISP's edge.

Network types, topologies and Wi-Fi standards from 802.11a to Wi-Fi 7

Networks are classified by geographic scope: PAN (personal area, Bluetooth, a few metres), LAN (local, a building or campus), CAN (campus, multiple buildings), MAN (metropolitan, city-scale, often fibre rings), WAN (wide area, MPLS, SD-WAN, the internet), SAN (storage area, Fibre Channel or iSCSI, a data centre fabric). The Indian National Knowledge Network is a MAN-to-WAN academic backbone connecting every IIT, IIIT, IISc and central university.

Topologies describe physical or logical arrangement. Bus (a single shared cable, the old 10BASE2 thinnet) is dead. Ring (Token Ring, FDDI) is largely dead though some industrial Modbus over fibre rings persist. Star (every host to a central switch) is the LAN default. Mesh (every node connects to multiple peers) appears in Wi-Fi mesh systems (Google Nest Wifi, TP-Link Deco), industrial 802.15.4 ZigBee networks, and at the SD-WAN layer where every site has tunnels to multiple peers. Tree or hierarchical is the classic three-tier enterprise design (access, distribution, core).

Architecture choices divide into client-server (a central host serves many clients, the model underlying HTTP, SMTP and IMAP) and peer-to-peer (every node both serves and consumes, the model underlying BitTorrent and the libp2p layer in IPFS). Peer-to-peer traffic is harder to attribute because there is no central server log to subpoena.

StandardMarketing nameYearBandMax link rate
802.11a(no name)19995 GHz54 Mbps
802.11b(no name)19992.4 GHz11 Mbps
802.11g(no name)20032.4 GHz54 Mbps
802.11nWi-Fi 420092.4 / 5 GHz600 Mbps (4x4 MIMO)
802.11acWi-Fi 520135 GHz6.9 Gbps (8x8, 160 MHz)
802.11axWi-Fi 620192.4 / 5 GHz9.6 Gbps (OFDMA)
802.11ax + 6 GHzWi-Fi 6E20206 GHz added9.6 Gbps
802.11beWi-Fi 720242.4 / 5 / 6 GHz46 Gbps (MLO, 320 MHz)

Wi-Fi 6E opens 1200 MHz of new spectrum in the 6 GHz band, approved for unlicensed use in India by the DoT in January 2026; that band hosts twice as many non-overlapping 80 MHz channels as the entire 5 GHz band did. Wi-Fi 7 adds Multi-Link Operation (MLO), where a single client maintains simultaneous associations on two bands and load-balances frames between them, and 320 MHz channels in 6 GHz. The forensic relevance is that a packet capture from a Wi-Fi 7 client requires the capture tool to follow MLO across radios, which only Wireshark 4.4+ with a recent libpcap and capable radio hardware can do.

The OSI seven-layer model with TCP/IP four-layer mapping. Each layer's example protocols and the artefact the digital forensi
The OSI seven-layer model with TCP/IP four-layer mapping. Each layer's example protocols and the artefact the digital forensic examiner recovers from a packet capture at that layer.

The CERT-In Direction dated 28 April 2022 mandates that data centres, virtual private server providers, VPN providers and cloud service providers operating in India maintain ICT system logs for 180 rolling days, register subscriber identification details, and synchronise all systems to the NPL or NIC NTP servers. Those logs are exactly the layer-by-layer artefacts above: DHCP leases mapping MAC to IP, NAT translation tables mapping internal port to external port, DNS query logs, and Layer 4 connection records. Every Indian network-forensic case after May 2022 starts with a CERT-In log request, and the examiner who does not know which header field lives at which layer cannot frame that request properly.

Practice
Question 1 of 5· 0 answered

A packet leaves an Indian ISP's NAT gateway and reaches a server in Frankfurt. Which header field is rewritten on every hop and therefore tells you nothing about the original device?

Frequently asked questions

What is the difference between the OSI model and the TCP/IP model?
The OSI model is a seven-layer reference framework (Physical, Data Link, Network, Transport, Session, Presentation, Application) published by ISO in 1984 for teaching and analysis. The TCP/IP model is the four-layer stack (Link, Internet, Transport, Application) defined by the IETF and actually deployed on the internet. The mapping is: OSI 1 and 2 collapse into TCP/IP Link; OSI 3 maps to TCP/IP Internet; OSI 4 to TCP/IP Transport; OSI 5, 6 and 7 collapse into TCP/IP Application. Forensic examiners use the OSI labels because Wireshark and most textbooks dissect by OSI layer.
Why does CIDR notation matter for digital forensics?
CIDR (`a.b.c.d/n`) is the only way to ask CERT-In or an ISP a precise question about which subscriber owned a given address at a given time. A log request worded as `203.0.113.0/24 between 02:15 and 02:45 IST on 14 May 2026` returns one customer; the same request worded loosely as `the 203.0.113.x range` returns ambiguity. CIDR also drives route aggregation in BGP, so understanding the prefix length lets the examiner interpret the path the traffic actually took.
What are RFC 1918 private addresses and why can't they appear on the public internet?
RFC 1918 reserves three IPv4 ranges for private use: 10.0.0.0/8, 172.16.0.0/12 and 192.168.0.0/16. Routers on the public internet are configured to drop packets with those source or destination addresses, so traffic between two private networks needs NAT at the edge that rewrites the private source IP into the organisation's public IP. The forensic consequence is that a packet capture from inside a private LAN shows different addresses than a capture upstream of NAT.
How does SLAAC work and what does it leak about the device?
Stateless Address Autoconfiguration (RFC 4862) is how an IPv6 host gets a global address without DHCPv6. The router multicasts a Router Advertisement with a /64 prefix; the host appends a 64-bit interface identifier. Old EUI-64 identifiers embedded the device MAC, leaking the OUI and serial. Modern hosts use RFC 7217 stable opaque identifiers or RFC 8981 temporary privacy identifiers that rotate every 24 hours, which strip that hardware fingerprint.
What is the practical difference between a hub, a switch and a router for an examiner?
A hub repeats every signal to every port (Layer 1, dead in 2026). A switch reads the Ethernet header and forwards a frame only to the port hosting the destination MAC (Layer 2). A router reads the IP header and forwards based on a routing table, rewriting MACs at each hop (Layer 3). Forensic relevance: a passive tap on a switched LAN sees only broadcast and the local port; a tap upstream of the router sees aggregated traffic from many devices but with rewritten MACs. The packet broker placement decides what the SOC can see.
Why does Wi-Fi 6E matter compared to Wi-Fi 6?
Wi-Fi 6E extends 802.11ax into the 6 GHz band, opening 1200 MHz of new unlicensed spectrum approved by the DoT in India in 2022. That band supports seven 160 MHz channels versus two in the entire 5 GHz band, so dense deployments (campus, stadium, conference) get far less congestion. For the examiner, Wi-Fi 6E means a capture target may be operating on 6 GHz channels that older sniffers (USB adapters limited to 2.4 and 5 GHz) cannot see at all.
What information about a device can I learn from its MAC address?
The first three octets (OUI) identify the vendor by an IEEE-assigned prefix; Wireshark resolves it automatically (`a4:83:e7` becomes Apple). The U/L bit of the first octet tells you whether the address is universally administered (vendor-assigned, 0) or locally administered (software-set, 1) and so reveals MAC randomisation in modern phones. The I/G bit distinguishes unicast (0) from multicast (1). Beyond that the lower three octets are the vendor's own serial and carry no public meaning, though some old vendors used predictable counters that let you estimate the manufacturing year.

Test yourself on Digital Forensics with free, timed mocks.

Practice Digital Forensics questions

Found this useful? Pass it along.

Share

Spotted an error in this page? Report a correction or read our editorial standards.

Your journey to becoming a forensic professional starts here.

Practice with mock tests, learn from structured notes, and get your questions answered by a global forensic community, all in one place.