Seeing the Big Picture

A system that identifies malicious patterns in network traffic could help create a more secure internet

Cancer cells, cilia development, air pollution, photonic devices, Micro-lens, mosquito-borne infections, Microbiota, bone repair, 3D printing, neurodegenerative disease, cancer treatments, biological research, sepsis, foot and mouth disease, cytometry, batteries, Influenza A virus, vascular diseases, New Cancer Drugs, RNA molecules, polymers, antimicrobial resistance, Aging White Blood Cells, microviscosity, Transplant Drug, Nanophotonics, photonics, Built-In Nanobulbs, cerebral cortex, cancer cells, nanowires, optoelectronic, solar energy, gold nanowires, Chikungunya virus, concrete, glaucoma, light-emitting diode, Proteomics, nanostructures, nickel catalyst, Ultrafast lasers, liver capsular macrophages, obesity, cancer, lignin polymer, liver capsular macrophages, Ultrafast lasers, monocyte cells, cancer treatments, antibody drug, gene mutations, quantum-entangled photons, gut microbes, skin aging, stroke, machine learning, Cloned tumors, cancer, Rare Skin Disease, terahertz lasers, silicon-nanostructure pixels, oral cancer, heart muscle cells, cancer, cancer stem cells, gastric cancer, microelectromechanical systems, data storage, silicon nanostructures, Drug delivery, cancer, muscle nuclei, Lithography, silicon nanostructures, Quantum matter, robust lattice structures, potassium ions, Photothermal therapy, Photonic devices, Optical Components, retina, allergy, immune cells, catalyst, Nanopositioning devices, mold templates, lung cancer, cytoskeletons, hepatitis b, cardiovascular disease, memory deficits, Photonics, pre-eclampsia treatment, hair loss, nanoparticles, mobile security, Fluid dynamics, MXene, Metal-assisted chemical etching, nanomedicine, Colorectal cancer, cancer therapy, liver inflammation, cancer treatment, Semiconductor lasers, zika virus, catalysts, stem cells, fetal immune system, genetic disease, liver cancer, cancer, liver cancer, RNA editing, obesity, Microcapsules, genetic disease, Piezoelectrics, cancer, magnesium alloy, Quantum materials, therapeutic antibodies, diabetes, 2D materials, lithium-ion batteries, obesity, lupus, surfactants, Sterilization, skin on chip, Magnetic Skyrmions, cyber-security, wound infections, human genetics, immune system, eczema, solar cells, Antimicrobials, joint disorder, genetics, cancer

The majority of cyber-security solutions that stand between us and increasingly sophisticated malware, target only specific attacks or subsets of attacks, meaning that users may have to buy and install many different products to protect themselves. Now, A*STAR researchers have developed a system that instead gathers evidence across a wide stream of internet traffic, and identifies links and correlations related to suspicious activity1.

“Our aim is to develop a framework to gather as much evidence as possible from a set of traffic, and indicate malicious anomalies, regardless of the type of attacks,” says Vrizlynn Thing at the A*STAR Institute for Infocomm Research, who led the study.

Thing and her team designed their new framework to look out for the fundamental characteristics of the malicious activities that stalk unsuspecting users through the evolving cyber landscape. Through this approach, the framework is robust against new threatening software and gathers only relevant evidence on the threats. For example, the system looks out for data flows that arrive at fixed time intervals, because attack bots are much less random than ordinary human-generated internet activities. The model also identifies sources that try to communicate with a large number of destinations in a short time, which is indicative of a botnet.

“The main challenge was devising ways to build up a large set of possible patterns which could serve as potential evidence for detecting a wide variety of anomalies,” says Thing. “We capture the persistent characteristics of the malicious activities in transit, and represent them in observable sequential forms. This has allowed us to detect very fundamental patterns related to malicious traffic.”

The team tested their new evidence-gathering system on recorded internet traffic, and found it could quickly identify many notorious botnets such as Andromeda, Zeus and Sality, with very few false positives. Given this success, Thing is hopeful that by improving their detection patterns, their system could defend networks against a much wider variety of attacks than has previously been possible.

“If we can detect malware infections by analyzing network traffic, we can prevent malware from further spreading,” she says. “We could also trigger the disconnection of infected hosts, thereby curbing the rampant growth of botnets.”

The A*STAR-affiliated researchers contributing to this research are from the Institute for Infocomm Research (I2R).

Source : A*STAR Research