SÉCURITÉ DES RÉSEAUX · CHEAT SHEET

Network Attacks Reference

All attacks extracted from course documents, Layers 1 through 7 · OSI Model · Attack name, description, symptoms, targeted layer, impact & protection.

Layers: L1 Physical L2 Data Link L3 Network L4 Transport L7 Application Severity: Critical / High Medium Low / Historical
LAYER 1 Physical Layer Attacks
Attack Name Description How to Detect (Symptoms) Layer Impact How to Protect
Electromagnetic Sniffing
EMI Eavesdropping / Van Eck
HIGH
An attacker uses a passive induction probe (Rogowski coil or close-range antenna) near a copper UTP cable to reconstruct data signals without any physical contact to the wire.
  • No direct network log, completely passive
  • Unusual personnel / equipment near cable runs
  • Electromagnetic anomalies detected by spectrum analyzers
L1 Confidential data exfiltration; undetectable by network tools Shielded cable (STP/FTP)Fiber opticPhysical access controlTEMPEST shieldingEnd-to-end encryption
Fiber Optic Tapping
Optical Bend Attack
HIGH
The attacker slightly bends a fiber optic cable to induce radiation loss from the core, then places an optical coupler to capture a fraction of the light signal without breaking the connection.
  • Slight signal attenuation increase (OTDR anomaly)
  • Unexplained light-level drops at monitoring points
  • Physical marks or clamps on fiber runs
L1 Silent data interception; no network-layer trace OTDR monitoringFiber conduit & locked ductsOptical power monitoringEnd-to-end encryption (MACsec/IPsec)
Unlocked Port Attack
Rogue Device Connection
CRITICAL
An attacker connects an unauthorized device (laptop, Raspberry Pi) to an unused, unlocked RJ45 port or an open network bay to sniff traffic, launch MiTM attacks, or pivot internally.
  • New MAC address appears on switch port
  • 802.1X authentication failure alert
  • Unusual traffic volume on an inactive port
  • DHCP request from unknown device
L1 Full network access; platform for MiTM, pivoting, data theft 802.1X port authentication (NAC)Port security (MAC limiting)Disable unused portsPhysical locks on baysCCTV / anti-tamper sensors
Console Port Attack
Evil Maid / Out-of-Band
CRITICAL
Physical access to UART/RS-232 or mini-USB console ports on routers/switches, which are typically unencrypted and require no network authentication, gives full device control.
  • Tamper-evident seals broken on equipment
  • Console cable detected on unauthorized port
  • Configuration changes without admin logins in audit log
L1 Complete device takeover; config modification; credential extraction Lock server rooms / network baysConsole port password + timeoutTamper-evident sealsTamper-detection switchesAnti-tamper screws
Supply Chain Firmware Attack
Hardware Backdoor
CRITICAL
Network equipment is delivered with pre-compromised firmware that can exfiltrate data, open backdoors, or disable security features from the moment of first power-on.
  • Unexpected outbound connections from new device
  • Firmware hash mismatch vs. vendor checksum
  • Unknown processes or ports active at boot
L1 Persistent undetectable backdoor; full device compromise from day zero Verify firmware hashesTrusted vendor sourcingSecure bootOut-of-band integrity checksNetwork baseline monitoring
Radio Jamming
RF Interference / DoS Wireless
HIGH
An attacker emits strong radio signals on the same frequency (2.4 GHz / 5 GHz) as a Wi-Fi network, preventing legitimate devices from communicating (physical-layer DoS).
  • Sudden drop in all wireless connectivity
  • Spectrum analyzer shows noise floor spike
  • Multiple clients disconnecting simultaneously
  • Signal strength normal but connection impossible
L1 Total wireless outage; can be used as cover for other attacks Frequency hopping (802.11n/ac)WIPS (Wireless IPS)Wired backup pathsRF shieldingSpectrum monitoring
LAYER 2 Data Link Layer Attacks
Attack Name Description How to Detect (Symptoms) Layer Impact How to Protect
ARP Spoofing
ARP Poisoning / MiTM
CRITICAL
The attacker broadcasts forged ARP replies, associating their MAC address with a legitimate IP (e.g., the gateway). Victims update their ARP cache and send traffic through the attacker, classic Man-in-the-Middle.
  • Two IPs mapped to the same MAC in ARP cache
  • arpwatch alerts on IP/MAC binding changes
  • Duplicate ARP reply packets in Wireshark
  • Unexpected gateway MAC change
L2 Traffic interception, credential theft, session hijacking, data manipulation Dynamic ARP Inspection (DAI)DHCP SnoopingStatic ARP entries (critical hosts)MACsec (802.1AE)HTTPS / TLS everywherearpwatch monitoring
MAC Flooding
CAM Table Overflow
HIGH
The attacker floods the switch with thousands of frames having random source MAC addresses (macof), exhausting the CAM table. The switch enters fail-open mode and broadcasts all frames on all ports, functioning as a hub.
  • show mac address-table count, abnormally high entries
  • SNMP: ifInErrors / ifInDiscards spike
  • All devices receive unicast traffic (promiscuous capture shows foreign frames)
  • Broadcast storm symptoms
L2 Full LAN traffic exposure; reconnaissance; enables further MiTM attacks Port Security (max MAC = 2)Violation shutdownStorm Control802.1X NACCAM table size limits
VLAN Hopping
Double Tagging Attack
HIGH
The attacker crafts frames with two 802.1Q VLAN tags (outer = native VLAN, inner = target VLAN). The switch strips the outer tag and forwards the frame into the target VLAN, bypassing segmentation without requiring a router.
  • Wireshark filter: vlan.id == 20 && vlan.id == 1 (double-tagged frames)
  • Traffic appearing in VLANs from unexpected source ports
  • IDS alert on double-encapsulated 802.1Q
L2 Unauthorized access to sensitive VLANs (admin, servers); VLAN isolation bypass Change native VLAN to unused VLAN (e.g. 999)switchport nonegotiateExplicit trunk VLAN allow lists802.1X on access ports
Rogue DHCP Server
DHCP Spoofing
HIGH
The attacker runs a fake DHCP server that responds to client DISCOVER requests faster than the legitimate server, handing out a malicious default gateway and DNS server, redirecting all traffic through the attacker.
  • DHCP OFFER from unexpected IP (Wireshark: bootp.option.dhcp == 2)
  • Cisco log: DHCP_SNOOPING_NONTRUSTPORT
  • Default gateway changed on client machines
  • DNS resolution failures or unexpected redirects
L2 Full MiTM position; DNS poisoning; credential interception DHCP Snooping (trust only uplink ports)IP Source GuardDAIDHCP lease monitoring802.1X
STP Hijacking
Rogue Root Bridge
CRITICAL
The attacker sends forged BPDUs with priority 0, claiming to be the Root Bridge. STP reconverges, making traffic flow through the attacker's device. Combined with MAC flooding, enables total LAN interception.
  • Wireshark: stp && (stp.root.prio == 0)
  • Cisco log: %SPANTREE-2-ROOTCHANGE
  • show spanning-tree vlan X, unexpected Root ID
  • Topology change notification storms
L2 Total traffic interception across all VLANs; network instability; loop risk BPDU Guard (all access ports)Root Guard (distribution ports)PortFast on end-host portsspanning-tree vlan X root primaryStorm Control
Wi-Fi Deauthentication
802.11 DoS / Evil Twin setup
HIGH
Attacker sends spoofed 802.11 deauthentication frames (which are unauthenticated in WPA2) forcing clients to disconnect. Used as DoS or to force reconnection to a rogue access point (Evil Twin).
  • Repeated client disconnections on all APs simultaneously
  • WIPS alert on deauth flood
  • Clients connecting to rogue AP with same SSID
  • Association/disassociation rate spike in AP logs
L2 Wireless DoS; credential capture via Evil Twin; WPA handshake capture for cracking 802.11w (Management Frame Protection)WPA3WIPS/WIDSCertificate-based auth (EAP-TLS)Rogue AP detection
LAYER 3 Network Layer Attacks
Attack Name Description How to Detect (Symptoms) Layer Impact How to Protect
IP Spoofing
Source Address Forgery
CRITICAL
The attacker forges the IP source address field (not validated by IP itself) to impersonate another host, bypass IP-based ACLs, hide identity, or as a vector for amplification DDoS attacks.
  • Traffic from impossible or reserved source IPs (RFC 1918 from WAN)
  • Asymmetric traffic patterns (packets in, no response route)
  • NetFlow showing same source IP for anomalous volume
  • Firewall logs: packets with internal source IP arriving on external interface
L3 ACL bypass; identity masking; foundation for DDoS amplification; non-repudiation failure BCP38 / uRPF (ingress filtering)ACL blocking bogon sourcesISP egress filteringIPsec authenticationNetFlow anomaly detection
ICMP Flood
Ping Flood / DoS
HIGH
The attacker sends a massive continuous stream of ICMP Echo Request packets (using hping3 or Scapy with inter=0.001) to saturate the target's bandwidth and exhaust CPU resources processing each request.
  • CPU spike on victim (top)
  • tcpdump -n icmp shows flood
  • Latency increase; packet loss on all services
  • Interface utilization at 100% (sar -n DEV)
L3 Service unavailability; CPU exhaustion; collateral impact on all hosted services ICMP rate limiting (iptables -m limit)Block unnecessary ICMP types (FW)Baseline alerting on ICMP volumeUpstream DDoS scrubbing
Smurf Attack
ICMP Amplification (Historical)
HISTORICAL
The attacker sends ICMP Echo Requests to a broadcast address with the victim's IP spoofed as source. All hosts on the broadcast network reply to the victim simultaneously, amplifying traffic by the number of hosts.
  • Massive inbound ICMP Echo Reply flood to victim
  • Traffic source appears as entire broadcast subnets
  • IDS signature: ICMP flood from multiple sources to single target
L3 Amplified DoS; network saturation; historically could take down ISP links Disable directed broadcast on routers (no ip directed-broadcast)BCP38 ingress filteringBlock ICMP at border FW
Ping of Death
Malformed Fragmentation (Historical)
HISTORICAL
The attacker sends IP fragments that, when reassembled, form an ICMP packet larger than 65535 bytes (max IP). Older unpatched OS kernels crash or reboot during the malformed reassembly.
  • Sudden system crash/reboot on target
  • Fragments with offset + length > 65535 in IDS logs
  • BSOD or kernel panic logs on victim
L3 System crash / reboot (DoS); no impact on patched modern OS OS patching (all modern OS immune)Firewall fragment inspectionIDS signature detection
Teardrop Attack
Overlapping Fragment DoS (Historical)
HISTORICAL
The attacker sends IP fragments with overlapping offset fields. When the target OS tries to reassemble the overlapping fragments, it encounters memory confusion and crashes on vulnerable (unpatched) systems.
  • System crash during reassembly of fragmented traffic
  • IDS alert: overlapping IP fragment offsets
  • Fragments with impossible offset values in packet capture
L3 System crash / kernel panic; DoS on legacy unpatched OS OS patchingDrop suspicious fragments at FWFragment reassembly limits
ICMP Redirect Attack
Routing MiTM via ICMP
HIGH
The attacker sends a forged ICMP Redirect message (Type 5) to a host, convincing it to use the attacker's machine as the "better gateway" for certain destinations, silently redirecting all matching traffic.
  • Host routing table shows unexpected gateway (route -n)
  • ICMP Type 5 messages from non-gateway sources in packet capture
  • Traffic going through unexpected next-hop
L3 Silent MiTM; traffic interception; complete session monitoring Disable ICMP Redirect acceptance on hosts (sysctl)Filter ICMP Redirect at FWStatic routes for critical pathsIPsec
ICMP Tunneling
Covert C2 / Data Exfiltration
HIGH
The attacker embeds arbitrary data (C2 commands, exfiltrated files) in the payload field of ICMP Echo packets. Most firewalls allow ICMP ping and don't inspect payloads, creating a covert channel.
  • ICMP packets with abnormally large payloads (>64 bytes)
  • Regular, periodic ICMP traffic (fixed interval suggests beaconing)
  • ICMP to/from external IPs not matching internal hosts
  • IDS: ICMP data pattern anomaly
L3 Data exfiltration bypassing FW; persistent C2 channel; undetected malware beaconing Deep Packet Inspection of ICMP payloadsRestrict ICMP to trusted hosts onlyNDR behavioral baselineNGFW application-layer inspection
RIP Route Injection
Routing Protocol Poisoning
HIGH
On networks running RIPv1 (no auth) or poorly configured RIPv2, the attacker (internal host or virtual router) announces a fake prefix with a better metric (e.g., hop count = 1). Routers install the poisoned route and forward traffic to the attacker.
  • RIP Response from unexpected source IP (UDP port 520)
  • Routing table shows unexpected next-hop (show ip route)
  • Sudden metric improvement for a prefix
  • NetFlow showing traffic leaving via unexpected interface
L3 Traffic redirection (MiTM); blackhole creation; inter-VLAN traffic interception RIPv2 MD5 authentication (all interfaces)Prefix lists (only accept expected prefixes)ACL on UDP/520 (allow only known peers)Migrate to OSPF/EIGRP with auth
BGP Prefix Hijacking
BGP Hijack / Route Theft
CRITICAL
A malicious or compromised AS announces a more specific prefix (e.g. /25 vs /24) that belongs to another AS. BGP best-path selection prefers longer prefixes, causing the internet to route victim traffic to the attacker's AS.
  • BGP monitoring tools (RIPE RIS, RouteViews) show prefix announced by wrong AS
  • Traceroute traverses unexpected ASes
  • Traffic drops to services hosted on hijacked prefix
  • BGP looking glass shows wrong origin AS
L3 Internet-scale traffic interception; service outage; certificate / BGP session hijacking RPKI / ROV (Route Origin Validation)IRR filteringBGP max-prefix limitsBGP MD5 / TCP-AO authenticationBGPsec (path validation)
BGP Route Leak
Route Propagation Violation
HIGH
An AS accidentally or maliciously re-announces routes learned from one peer to another peer, violating routing policies. Can cause mass traffic redirection through an unprepared AS (sometimes causing Internet outages).
  • Sudden increase in BGP prefix count from a peer
  • Traffic from many ISPs routes through unexpected AS
  • BGP table shows transit routes from a stub network
L3 Regional or global traffic rerouting; latency spikes; potential interception Explicit BGP export policiesNo re-announce by defaultMax-prefix session limitsRPKI ROVIRR registration
OSPF LSA Falsification
Link-State Poisoning
HIGH
Without OSPF authentication (or with weak MD5), an attacker sends forged LSAs (Link State Advertisements) claiming fake topology, creating blackholes, redirect paths, or routing instability across the OSPF area.
  • OSPF log: LSA sequence number anomalies
  • Unexplained SPF recalculations
  • Routes appearing/disappearing rapidly (flapping)
  • OSPF neighbor adjacency resets
L3 Routing blackholes; traffic redirection; network instability; DoS OSPF MD5 / SHA authentication (all interfaces)Passive interfaces on non-router portsOSPF area segmentationSIEM alerting on LSA anomalies
LAYER 4 Transport Layer Attacks
Attack Name Description How to Detect (Symptoms) Layer Impact How to Protect
TCP SYN Flood
Half-Open Connection Exhaustion
CRITICAL
The attacker sends massive numbers of TCP SYN packets (often with spoofed source IPs) without completing the 3-way handshake. The server allocates a TCB (half-open connection) for each SYN, eventually exhausting its connection table.
  • Large number of SYN_RECV connections in ss -s
  • Server connection table at max (netstat overflow)
  • Legitimate users cannot establish TCP connections
  • High SYN packet rate in NetFlow/IDS
L4 Service unavailability; server resource exhaustion; all TCP services affected SYN cookies (RFC 4987)Rate limiting SYN packets (iptables)Firewall state table limitsSCTP (cookie-based handshake)Upstream DDoS scrubbing
Session Hijacking
TCP Session Takeover
CRITICAL
After passively sniffing or predicting TCP sequence numbers, the attacker injects forged packets with the correct sequence numbers into an active session, taking control away from the legitimate client.
  • Duplicate ACK / unexpected RST in existing connections
  • Session "stuck" or reset mid-transaction
  • Authentication actions performed without user interaction
  • IDS: TCP sequence anomaly alerts
L4 Unauthorized actions on behalf of authenticated user; data theft; account compromise Encrypted sessions (TLS 1.3)Random initial sequence numbers (OS-level)Short session timeoutsApplication-level re-authentication
TCP RST Attack
Connection Reset Injection
HIGH
The attacker sends forged TCP RST (reset) packets with the correct sequence number to an active connection. Both endpoints immediately terminate the connection, believing the other side requested it.
  • Sudden unexpected connection drops
  • RST packets from IPs not matching the actual peer
  • Applications reconnecting repeatedly
  • IDS: RST injection pattern
L4 Session disruption; denial of service for long-lived connections (BGP, VPN); transaction loss TCP-AO (Authentication Option, RFC 5925)IPsec ESP for session protectionFirewall stateful trackingTLS session continuity
UDP Amplification / Flood
Reflection DDoS (DNS, NTP, SSDP)
CRITICAL
Using IP spoofing, the attacker sends small requests to UDP services (DNS, NTP, Memcached) with the victim's IP as source. The servers send much larger responses to the victim, amplification factor can exceed 50,000x for Memcached.
  • Massive inbound UDP traffic from many distributed IPs
  • Traffic spike on DNS/NTP/SSDP ports toward victim
  • Origin IPs are legitimate public servers (amplifiers)
  • Bandwidth saturation at upstream link
L4 Multi-Gbps DDoS; complete network saturation; collateral damage to ISP infrastructure BCP38 to prevent IP spoofing at sourceRate-limit UDP responses (server-side)Disable open recursive DNS / public NTPUpstream DDoS mitigation (Akamai, Cloudflare)
Port Exhaustion
Ephemeral Port Starvation
MEDIUM
The attacker rapidly opens thousands of connections to the target, consuming all available ephemeral ports (typically 1024-65535) on the server or NAT gateway, preventing any new connections from legitimate clients.
  • Connection failures with "address already in use" errors
  • All ephemeral ports in TIME_WAIT state (ss -s)
  • New connection attempts time out
  • Port count at OS maximum
L4 Service unavailability for new connections; effective DoS without flooding Connection limits per source IP (FW)Reduce TIME_WAIT timeoutSO_REUSEPORT socket optionRate limiting per client
Fragment Overlap Attack
IDS Evasion via Fragmentation
HIGH
The attacker sends packets split into small fragments that individually appear benign, but when reassembled by the target contain malicious content that was hidden from IDS/IPS which only inspects fragments individually.
  • Many small UDP/ICMP fragments from same source
  • IDS low-fragment alerts followed by application-layer attacks
  • Exploit traffic that bypassed perimeter IDS
L4 IDS/IPS bypass; delivery of exploits that should have been blocked NGFW full-reassembly inspection before processingDrop fragments below minimum sizeFragment timeout limitsNormalize fragmented traffic
LAYER 7+ Application Layer & Multi-Stage Attacks
Attack Name Description How to Detect (Symptoms) Layer Impact How to Protect
SQL Injection
SQLi / DB Exfiltration
CRITICAL
Attacker injects SQL code into input fields (e.g., admin' OR '1'='1' -- in login forms). The application passes the malicious query to the database, potentially bypassing authentication or exposing all data.
  • IDS: SQL keywords in HTTP POST body (OR, UNION, DROP)
  • Unusual database query patterns in DB logs
  • Login without valid credentials
  • Error messages exposing DB structure
L7 Full database compromise; authentication bypass; data exfiltration; initial foothold in APT Parameterized queries / prepared statementsWAF (Web Application Firewall)Input validation & sanitizationLeast-privilege DB accountsSIEM alerting on SQL patterns
DNS Tunneling
DNS Covert Channel / Exfiltration
HIGH
Data is encoded into DNS query/response fields (TXT, CNAME records). Since DNS is rarely blocked or inspected, this creates a covert bidirectional channel for C2 communication or data exfiltration through corporate firewalls.
  • Abnormally frequent DNS TXT queries (every 3s in APT scenario)
  • Long, base64-like subdomains in DNS queries
  • DNS queries to rare/newly-registered domains
  • Volume of DNS traffic far above baseline
  • DNS responses with large TXT records
L7 Data exfiltration bypassing DLP; persistent C2 channel; undetected over long periods DNS query rate limitingInspect TXT record content (NGFW)DNS RPZ (Response Policy Zones)Suricata/Sigma DNS tunneling rulesDNS traffic baseline anomaly detection
Pass-the-Hash
Credential Reuse / Lateral Movement
CRITICAL
After compromising one machine, the attacker extracts NTLM password hashes from memory (without knowing the plaintext). These hashes are directly used to authenticate to other Windows systems via SMB/WMI, no cracking needed.
  • SIEM: multiple auth attempts from a single internal IP to many hosts
  • Event ID 4624 (Logon Type 3/Network) from unusual sources
  • Lateral movement pattern: same hash used across subnets
  • EDR: credential access on LSASS process
L7 Rapid lateral movement; privilege escalation; domain compromise Windows Credential GuardDisable NTLM (enforce Kerberos)Tiered Admin Model (PAW)EDR on all endpointsNetwork segmentation / micro-seg
Phishing + Macro Attack
Spear Phishing / VBA Payload
CRITICAL
Attacker sends a crafted email with a malicious Office document containing a VBA macro. When the user enables macros, the macro downloads and executes a payload via HTTPS (port 443), establishing a C2 channel that bypasses firewall inspection.
  • Email with Office attachment from unexpected sender
  • Office process spawning cmd.exe or PowerShell as child process (EDR)
  • Outbound HTTPS to newly-registered or low-reputation domain
  • Antivirus/EDR: macro execution alert
L7 Initial access in APT kill chain; C2 establishment; ransomware delivery; credential theft Disable macros by default (GPO)Email filtering (sandbox detonation)NGFW TLS inspection (port 443)User awareness trainingEDR behavioral detection
Ransomware (Multi-Stage)
APT Ransomware / Wiper
CRITICAL
Following initial access (phishing), the attacker moves laterally via SMB/RDP with stolen credentials, then deploys ransomware that encrypts files on all reachable file shares. All stages use legitimate protocols to evade detection.
  • Sudden mass file extension changes (.locked, .enc)
  • Ransom note files appearing on shares
  • SMB traffic spike from single source to many hosts
  • EDR: process injecting into svchost.exe
  • C2 beacon to external IP (185.x.x.x range)
L7 Complete data loss / encryption; business disruption; regulatory consequences; multi-million € ransom Immutable offline backups (3-2-1 rule)EDR + anti-ransomware moduleNetwork micro-segmentation (block SMB spread)Disable RDP or require MFAZero Trust architectureIncident Response Plan
APT, SQL → PtH → SCADA
Advanced Persistent Threat Kill Chain
CRITICAL
Full multi-stage APT (as in the SOC scenario): SQL injection on DMZ web app → brute-force internal service account → Pass-the-Hash lateral movement → pivot to OT/SCADA via misconfigured IT/OT firewall → write SCADA register to open circuit breaker → DNS tunneling for data exfiltration.
  • 03:17, IDS: SQL injection in POST body
  • 03:19, 47 auth failures on svc_backup (AD log)
  • 03:22, SIEM: lateral movement (SMB, Pass-the-Hash)
  • 03:31, FW: unauthorized IT→OT connection (port 102 S7)
  • 03:38, SCADA: unplanned write register command
  • 03:44, DNS: TXT queries to evil-domain.xyz every 3s
L3-L7 Physical infrastructure damage (power outage); mass data exfiltration; regulatory / safety consequences WAF on DMZ web appsMFA on all accountsPatch AD (6-month lag = critical)Strict IT/OT firewall rules (no inherit)Encrypted S7comm (TLS)EDR on all IT endpointsNDR + SIEM correlationSOAR auto-isolation playbooks