Too many big data initiatives are science projects that take months of effort, risk failure and require highly trained data scientists with scarce skills. According to a CSC survey, 55 percent of big data projects aren’t completed and many others fall short of their objectives.Read more...
‘Big Spectrum Data’ Could Help Predict Wireless Net Usage
The latest on the growing list of big data applications is the prospect of leveraging data analytics to predict the utilization of wireless networks.
A recent paper by researchers in China and the U.S. titled Big Data Spectrum: The New Resource for Cognitive Wireless Networkingdescribed how big data analytics can be used to measure how wireless spectrum is being used. The researchers also offered guides to developing algorithms needed to track wireless network utilization.
They argued that the need to understand network utilization is growing as spectrum grows more congested and the total number of mobile subscribers reaches 4 billion. Add to that, they estimated that about 30 billion radio-frequency identification (RFID) tags were produced globally over the last two years.
RFID tags are widely used in global supply chains to track inventory. The tags are slowing being incorporated into an emerging “Internet of Things” that links a range of wireless devices.
“All these sensors generate data,” the researchers stressed. “Sadly, much of this data is simply thrown away because of the lack of efficient mechanisms to derive value from it.”
Radio spectrum used for mobile networks is not physically scarce but tends to be vastly underutilized. Based on temporal and geographic variations, utilization rates vary widely from between 15 percent to 85 percent.
To improve radio spectrum utilization, cognitive wireless networking is seen as a promising model, providing dynamic spectrum sharing among heterogeneous networks. This requires the creation of a “geolocation spectrum database” along with “radio environment” maps that depend on efficient techniques to sift through complex spectrum data sets.
Hence, the researchers argue, “big spectrum” data analytics can be applied to cognitive wireless networking to improve radio spectrum utilization.
Spectrum data refers to all data related to spectrum state (whether idle, in use, signal energy levels, and signal features) as well as user and device data such as device ID. It also includes “environment side data” such as terrain, meteorological, and hydrographic information.
Meanwhile, the researchers defined big spectrum data as specific patterns in big data in the wireless domain and other spectrum data too complex to be analyzed using traditional methods.
Big spectrum data is further divided by volume, value (spectrum modeling, prediction), variety, velocity (data in motion), veracity (quality of data) and viability.
Volume tends to be a key characteristic of big spectrum data, the researchers said. “The sheer volume of spectrum data being stored and processed is exploding at an unprecedented rate, which is mainly driven by the need to gain a full understanding of radio spectrum dynamics.”
The authors also argue that big spectrum data “allows a more complete picture of radio spectrum usage and deeper understanding of the hidden patterns behind spectrum state evolution and spectrum utilization.”
The velocity of big spectrum data refers to the fact that spectrum data is being generated at a rate greater than what traditional systems can handle. Hence, the need for big data analytical tools.
The researchers recommended a series of research initiatives to speed deployment of big spectrum data analytics for use in predicting wireless utilization.
“Solutions to the challenges ahead will require ideas from computer science and statistics, with essential contributions also needed from applied mathematics, from optimization theory, and from various engineering areas, notably wireless networking, signal processing and information theory,” they concluded.