A network service that traps more malicious software than a single antivirus program catches by itself could be the next weapon used to fight Internet threats.

Researchers from the University of Michigan, who developed the CloudAV service, contend that antivirus programs don't detect a substantial percentage of malware. Also, they say there's a time lag between when a threat appears and when the antivirus program is updated to detect it.

Security experts warn that people should use antivirus products, but also the programs' effectiveness is slowly diminishing with an ever-increasing rise in malicious software.

The researchers' method uses the "cloud-computing" concept, where the processing of a task is performed on a remote server and the result is delivered back to a PC or a mobile device.

[ Learn more about what cloud computing really means and the new breed of utility computing and platform-as-a-service offerings. ]

CloudAV uses a muscular approach, combining 10 antivirus engines and two behavioral detection ones into one service. The researchers took a cue from "N-version programming," a method in which different software implementations are used to ensure the reliability of services such as file systems.

"Antivirus engines have complementary detection capabilities, and a combination of many different engines can improve the overall identification of malicious and unwanted software," according to CloudAV. "This model enables identification of malicious and unwanted software by multiple, heterogeneous detection engines in parallel, a technique we term N-version protection."

To use CloudAV, a host agent is installed on a PC running either Windows, Linux or the FreeBSD operating systems. The agent can also be installed on a mobile device.

The agent monitors new files and programs that are written to disk. A cache is created of previously analyzed files to reduce load on the network. New files not recognized in the local cache are sent to the network. CloudAV can compare it with its cache or run an analysis, which takes around 1.3 seconds.

During six months of testing, CloudAV detected 98 percent of some 7,220 malware samples researchers ran against it. A single detection engine only gets 83 percent, the researchers wrote.

The antivirus engines used by CloudAV are Avast, AVG, BitDefender, ClamAV, F-Prot, F-Secure, Kaspersky, McAfee, Symantec, and Trend Micro -- plus two behavioral detection engines, Norman Solutions' Sandbox and Sunbelt Software's CWSandbox.

The researchers caution that network services such as CloudAV won't replace antivirus or intrusion-detection software, but could be used in combination to create a better defense against malware.

The research paper was authored by Jon Oberheide, Evan Cooke and Farnam Jahanian of the Electrical Engineering and Computer Science Department at the University of Michigan.