HOME

RESEARCH

PUBLICATION

TEACHING

ALBUM

 

**The following figures and text on this page are under GUN Free Documentation license unless specially noted :)**

My recent research contribution has been on the architecture of large-scale cognitive wireless networks. The technology differentiates from state-of-the-art, by its opportunistic network resource utilization of both spectrum bandwidth and mesh station/radio availability. On the contrary, state-of-the-art wireless networking assumes that those resources can be predetermined, by importing the protocol stack from cable networks. For example, in the traditional protocol stack, MAC (Media Access Control) layer allocates spectrum resources to wireless linkage; and network layer sets up routing path from source to destination based on established topology. In large-scale wireless systems, the basic network architecture introduces bottlenecks along both wireless links and stations: volatile spectrum availability is typical in unlicensed bands where interference prevails; and random station/radio availability is also often encountered due to the dynamic traffic load and other factors such as radio failure.

By taking the cross-layer architecture “Embedded Wireless Interconnect” (outlined in the figures below) that merges network routing into wireless link and RF design, the investigated technology creates a dynamic (fluid) wireless network without predetermined topology and spectrum allocation. For example, in multi-hop wireless communications, every packet takes opportunistically available paths in the wireless network, and with opportunistically available spectrum on each hop. The network-resource utilization can thereby reach its instantaneous maximum, disregarding volatile changes and the demand placed on the network.

 

 

As an analogy, given a data packet as a “car”, traditional wireless networking is analogous to driving the car according to a static roadmap; whereas the large-scale cognitive networking is analogous to driving with the guidance of a smart GPS integrated with real-time traffic information.

With regards to network-scalability specifically, state-of-the-art technologies suffer from 1) complexity increases fast with network scale; 2) performance decreases fast with network scale. Usually both can be getting worse exponentially in implementations due to the complexity of maintaining routing paths and routing tables. For our developed cognitive-networking technology: 1) complexity can be constant with network scale due to instantaneous local observations; 2) performance can increase with network scale due to more alternative paths to exploit. It makes cheap large-scale wireless networks feasible, while state-of-the-art technologies need bulky and expensive boxes. The ultimate implementation can be a single-chip wireless mesh solution as the network can be best integrated in silicon chip. This can bring down the node cost to $1-5, and enjoy much superior performance than the current >$1000 systems.

In essence, it provides the best cost-effective bandwidth in large-scale wireless infrastructures. Compared to state-of-the-art competing technologies, it has been technically and experimentally proved to achieve 5-10 times higher throughput (bandwidth) in wireless networks, with a fraction of costs in terms of materials, installation, and maintenance. As large wireless infrastructures have always been expensive to build and maintain (e.g., cellular towers), wireless bandwidth has been expensive to attain. The full commercialization will potentially be a game-changer that can bring the best cost-effective wireless connectivity to “everybody” and “everything”. “Everybody” can cover current telecommunications to address the pain of insufficient bandwidth and coverage of wireless Internet as consumed by smart phones; and “everything” can bring smart wireless connectivity to sensor-devices in many traditional industries, including smart grid and water networks, mining, healthcare, surveillance, emergency communications, agriculture, home/building automation, indoor location networks, and retailers.

As such, the investigation is a platform with enormous potentials in vertical applications. It will ultimately make wireless communications scalable and affordable by reaching ubiquitous, where the full impacts can be comparable to: 1) what packet-switch network technology (Internet) has brought to personal communications; 2) what mobile communication technology (cellular) has brought to telephony.

Networking Method

Propagation Medium

Network Traffic

Circuit Switch

Reliable

Predetermined

Packet Switch

Reliable

Random

Investigated Technology

Unreliable

Random

 

Here is also a list of technology merits that can be applied to a broad set of applications:

·         Support scalable wireless networks with high performance;

·         Robust to external interferences and harsh networking environment;

·         Support high mobility with reliable communications;

·         Low complexity, low power and low cost.

Here are further a few simulated results for technology comparison in interference intensive (unlicensed) and large scale (multi-hop) networking environments:

 

说明: 说明: Drawing2 说明: 说明: Drawing1

 

* Note: the legend "OMESH" in the above figures (© OMESH Networks Inc.) denotes the investigated large-scale cognitive networking technology.

** OMESH Networks Inc. (Toronto, Canada) is supplying cognitive radios integrated with the technology.

 

 

 

 

[ home | research | publication | teaching | album ]

© Liang Song  2011.