Memories for the Intelligent Internet of Things

6Big Data Search Engines and Deep Computers

6.1 Overview of Big Data Search Engines and Deep Computers

The Internet of Things (IoT) collects a large amount of data, much of which is stored in large server farms called collectively “The Cloud”. These data are only useful if they can be accessed intelligently and efficiently. An example of a basic data search engine is a ternary content addressable memory (TCAM). This is a memory that can compare input search data with stored data and return the location of matched data. TCAMs are used in both MPU designs and in communication chips such as those for IP‐routing.

Deep computers and artificial intelligence will also be needed to intelligently process the data in The Cloud. In addition to neuromorphic computers, which were discussed in an earlier chapter, extreme learning machine architectures and artificial intelligence are being developed. Deep learning can be implemented using a restricted Boltzman machine, which consists of a generative stochastic artificial neural network. Deep learning involves extracting complex information from high‐dimensional data without using significant amounts of manual engineering. It is useful for image recognition, speech recognition, and natural language understanding.

6.2 Content Addressable Memories Made Using Various Emerging Nonvolatile Memories

A content addressable memory, or associative memory, searches its entire contents in a single clock cycle. It is commonly used for high speed search applications. ...