Transcript forth_wisdom_crete09

Workpackage 3
WISDOM
New security algorithm design
ICS-FORTH
Heraklion, 3rd June 2009
WISDOM WP3: New security
algorithm design
WISDOM
Objectives
• Identify critical security application components which
can be efficiently implemented in the optical domain
• Characterise constraints to algorithmic components and
develop novel techniques for simplified pattern matching
• Design a Security Application Programming Interface
(SAPI) which will be the interface between high-level
security applications and low-level optical
implementation
Tasks – Deliverables
• WP3.1: Security Applications Partitioning (M12)
• WP3.2: Identification of Simplified Security Algorithm
Components (M24)
• WP3.3: Definition of a Security Application Programming
Interface: SAPI (M30)
WP3.1 Security Applications
Partitioning
WISDOM
Identify efficient operations in
optical domain by considering
• basic firewall functionality
prevents communication for
specific servers and services
• basic IDS/IPS functionality
signature, anomaly based detection
4
• packet structure and decoding
TOS
16-bit identification
TTL
header (fixed length)
payload (variable length)
protocol
16-bit total length
flags
13-bit fragment offset
16-bit header checksum
32-bit source IP address
32-bit destination IP address
options (if any)
• optical hardware
optical data format, optical bit filtering,
optical pattern matching,
buffer (delays)
IHL
16-bit source port
16-bit destination port
32-bit sequence number
32-bit acknowledgment number
Offset
Reserved
Flags
16-bit checksum
16-bit window
urgent pointer
Options (if any)
Application data
WP3.1 Security Applications
Partitioning
WISDOM
Critical security operations in the optical domain
Basic firewall functionality, inspect packet headers
Less than 10% of rules, more than 90% of alerts
Look at specific packet header field
• Block or filter traffic for specific protocols, ports, etc
Optical filtering, optical pattern matching, optical routing
• Block or filter traffic for specific IP addresses
Optical possible but not efficient
Combined inspections of several header fields
• Specific IP protocols and ports
• From specific IP addresses to specific ports
Optical possible but combination of optical and electronic more
efficient
WP3.1 Security Applications
Partitioning
Firewall rule example
WISDOM
Inspection
•
•
•
•
•
•
Deny all incoming traffic with IP matching internal IP
source IP address
Deny incoming from black-listed IP addresses
source IP address
Deny all incoming ICMP traffic
IP protocol
Deny incoming TCP/UDP 135/445 (RPC, Windows Sharing) destination port
Deny incoming/outgoing TCP 6666/6667 (IRC)
destination port
Allow incoming TCP 80, 443 (http, https)
destination port
to internal web server
(destination IP address)
• Deny incoming TCP 25 to SMTP server
destination port
from external IP addresses
(destination)/source IP address
• Allow UDP 53 to internal
destination port
DNS server
(destination IP address)
typical port assignments for some other services/applications
ftp TCP 21, ssh TCP 22, telnet TCP 23, POP3 TCP 110, IMAP 143
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
• Optical pre-processing for more ‘traditional’ IDS
– Restrictions in optical domain (buffering, level of integration, etc)
– Scalability of security pattern matching algorithms, optimum
balance between optical and electronic processing (WP6)
– Develop algorithms that will allow optical bit-serial processing
subsystems to operate as a pre-processor to more complex
pattern recognition techniques in the electronic domain.
D3.2 Identification of Simplified Security Algorithms
Components
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
• Identify feasible and efficient all-optical operations
– Inspection of specific fields in packet headers (protocol number,
port number, etc)
– Pattern matching
– Routing
• Keep all options for conventional (electronic) IDS
– Design high speed optical pre-processing that makes electronic
processing more efficient
• Demonstration of key security functions
– Example applications with efficient and reliable operation of a
hybrid system consisting of both all-optical and electronic
components
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
Combine optical and electronic signature-based detection
• Optical traffic splitters
– optical header processing
– split high speed network traffic
– group packets, e.g., according to port number
• Multiple “specialized” (electronic) processors
– less packets to inspect per processor
– more efficient payload inspection by performing same operations
to same type of packets
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
Approach for Hybrid Optical – Electrical Platform
• All-optical inspection of packet headers only
• A few well chosen useful rules optically implemented
– Restrictions in memory and level of integration imply small
number of selected rules can be implemented in optical domain
– Reconfigurable optical systems
– Analysis and statistics of network security threats
• Seamless coupling of optics with electronics
– Electronic processing enhanced by optical preprocessing
– Security applications (including payload inspection) in electronic
domain with more conventional NIDS tools
– Take advantage of “conventional” NIDS/NIPS methods
continuously developed
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
Use network traffic monitoring and classification appmon
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
Select rules using statistics on suspect packets
NoAH honeypots statistics
Protocols
Ports
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
• Network traffic monitoring
– Deployment of network of sensors for
global view
• Protocols
– ICMP often used in attacks
– TCP most popular, UDP also heavily used
• Ports
HEAnet
– Some high level applications use TCP/IP with pre-assigned port
numbers
– Others use dynamically assigned port numbers, different for
different connections
– Some attacks work on specific ports
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
Benefits from optical splitting for electronic processing
Similar approaches already proved successful in intensive
NIDS applications
• Early filtering and forwarding
• Packets of the same type are grouped by the splitter and forwarded
to specialized electronic processors
• Performance benefits (50-90/%) with the use of digital network
processors
• Clustering of packets with same destination port number improves
performance of conventional IDS
40% increase in packet processing throughput
60% improvement in packet loss rate
WP3.2 Identification of
Simplified Security Algorithms
Components
Available hybrid integrated optical circuits:
• XOR, AND logic gates
• buffer memory (limited)
• routing switch
• Bit pattern matching circuit
• Target pattern generator
• Pseudo random bit sequence generator
• Header sampler (proposal)
• CRC (proposal)
WISDOM
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
Input: flux of packets, consisting of RZ pulses T
Output: packets dropped or allowed to proceed
Preamble
Box: Header sampler
Header
Buffer memory
Header
Header
Header
Payload
Guard band
Preamble
MZI1
Preamble
Preamble
Bit pattern matching
Routing switch
TCP Port #
TCP Port #
Payload
Guard band
Header
TCP Port #
Header
Payload
Guard band
Header
TCP Port #
Header
Payload
Guard band
CRC
WP3.2 Identification of
Simplified Security Algorithms
Components
Same components, simple pipelined configuration
8 bit pattern matching at left boxes
16 bit pattern matching at right boxes
Possible packet collisions, bottleneck
WISDOM
WP3.2 Identification of
Simplified Security Algorithms
Components
“router”:
round-robin, CRC
WISDOM
WP3.2 Identification of
Simplified Security Algorithms
Components
WISDOM
Simulator of optical device operation
Basic building blocks are logic gates
Useful for circuit design, testing efficiency of proposed configurations,
analysis of more complex algorithms, hybrid optical-electronic detection,
load balancing, parallel/distributed configurations, anomaly-based
detection, etc.
WP 3.3 Definition of a Security
Application Programming
Interface (SAPI)
WISDOM
• Software platform – “mini” operating system
bridges the gap between optical execution of key components and
programming of security applications
• High-level programming, abstract all low-level details
operates independent of system modifications, allows for integration
of additional software and hardware components of increasing
complexity
• Hardware – software interface
fast optical processing, reconfigurable at much slower rates
user interventions rare, at conventional speed of electronics
• Front-end for SAPI and WSIM
Hardware and simulator run under same environment
D3.3 Definition of SAPI (M30)
WP 3.3 Definition of a Security
Application Programming
Interface (SAPI)
WISDOM
• Device configuration
hardware control
• Set security application rules
predefined filters
custom filters
• System monitoring
visualization of security operations outcome
• Easy to use GUI at front-end
user friendly control panel
same for actual operation and simulation
• Testbed and more complex systems
designed to support any hybrid optical-electronic architecture
WP3: SAPI
WISDOM
Interface with hardware and simulator (details coming up)
WP3: SAPI
WSIM tool
WISDOM
WP3: SAPI
WSIM tool
WISDOM
…more to follow…
WP3: New security algorithm
design
WISDOM
• Basic Firewall functionality in the optical domain (D3.1)
– Feasible, useful, and efficient packet header fields inspection
• Optical pre-processing for electronic NIDS/NIPS (D3.2)
– Actual security threats taken into account through network
monitoring and attack statistics
– All-optical header inspection and packet classification combined
with electronic processing of payload
– Proposal for hybrid systems with optimum balance between
optical and electrical processing:
optical enhances electrical, benefits from conventional electronic
NIDS/NIPS preserved
• SAPI (D3.3)
– High-level programming of security applications running over
optical and electronic hardware
• Functional optical device simulator
–
–
Complex algorithm design
Development that may be of more general interest
WP3 concluded: What next?
WISDOM
• Prepare SAPI for (upcoming) demonstrator/hardware
• Quantify benefits from optical pre-processing
Extensive processing of actual traces with simulator
Test different scenarios, DoS attacks, etc
Constant improvement of simulation tools
• Include physical models of optical devices in simulations
Perhaps not essential to this project, but overall very important…
• Details on high performance commercial NIDS
Endace, Crossbeam, etc., simple parallelization, dumb load balancers,
or more?
Convince about advantages of all-optical pre-processing
• Future
‘Green’ aspects of project (e.g., low power consumption)
Think again about payload inspection (partially in optical domain)
What is feasible in terms of optical components and devices?