Asad I. Khan

A Hierarchical Graph Neuron Scheme for Real-Time Pattern Recognition

The hierarchical graph neuron (HGN) implements a single cycle memorization and recall operation through a novel algorithmic design. The HGN is an improvement on the already published original graph neuron (GN) algorithm. In this improved approach, it recognizes incomplete/noisy patterns. It also resolves the crosstalk problem, which is identified in the previous publications, within closely matched patterns. To accomplish this, the HGN links multiple GN networks for filtering noise and crosstalk out of pattern data inputs. Intrinsically, the HGN is a lightweight in-network processing algorithm which does not require expensive floating point computations; hence, it is very suitable for real-time applications and tiny devices such as the wireless sensor networks. This paper describes that the HGNs pattern matching capability and the small response time remain insensitive to the increases in the number of stored patterns. Moreover, the HGN does not require definition of rules or setting of thresholds by the operator to achieve the desired results nor does it require heuristics entailing iterative operations for memorization and recall of patterns.

Parallel pattern recognition computations within a wireless sensor network

The computational properties of a wireless sensor network (WSN) have been investigated by implementing a fully distributed pattern recognition algorithm within the network. It is shown that the set up allows a physical object to develop a capability, which to some extent may be considered similar to our sense of touch, with the WSN acting as an artificial nervous system in this regard. The effectiveness of the algorithm is inspected by comparing the outputs from the sensors with the stress patterns generated through a simple finite element model and then stored within the network. It is shown that the test object could successfully differentiate between its internal stress states resulting from the changes to its external loading conditions. Suitability of the algorithm is discussed with respect to the data storage requirement per node of the WSN.

A parallel distributed application of the wireless sensor network

The paper describes the use of a wireless sensor network (WSN) for performing parallel pattern recognition computations. A complexity analysis indicates that the proposed algorithm is independent of the number of nodes and hence may scale up indefinitely with the network. Its shown that any material object once overlaid with a WSN, develops a latent associative memory, which enables the object to memorise some of its critical internal states for a real time comparison with those induced by the transient external conditions.

One shot associative memory method for distorted pattern recognition

In this paper, we present a novel associative memory approach for pattern recognition termed as Distributed Hierarchical Graph Neuron (DHGN). DHGN is a scalable, distributed, and one-shot learning pattern recognition algorithm which uses graph representations for pattern matching without increasing the computation complexity of the algorithm. We have successfully tested this algorithm for character patterns with structural and random distortions. The pattern recognition process is completed in one-shot and within a fixed number of steps.

A Pattern Recognition Scheme for Distributed Denial of Service (DDoS) Attacks in Wireless Sensor Networks

We define distinct attack patterns depicting distributed denial of service (DDoS) attacks against target nodes within wireless sensor networks for three most commonly used network topologies. We propose a graph neuron (GN)-based, decentralized pattern recognition scheme for attack detection. The scheme does analysis of internal traffic flow of the network for DDoS attack patterns. We stipulate that the attack patterns depend on both the current energy levels, as well as the energy consumption rates of individual target nodes. The results of varying pattern update rates on the pattern recognition accuracies for the three network topologies are included in the end to test the effectiveness of our implementation

Implementing a graph neuron array for pattern recognition within unstructured wireless sensor networks

Graph Neuron (GN) is a network-centric algorithm which envisages a stable and structured network of tiny devices as the platform for parallel distributed pattern recognition. However, the unstructured and often dynamic topology of a wireless sensor network (WSN) does not allow deployment of such applications. In this paper, using GN as a test-bed application, we show that a simple virtual topology overlay would enable distributed applications requiring stable structured networks to be deployed over dynamic unstructured networks without any alteration.

Single-Cycle Image Recognition Using an Adaptive Granularity Associative Memory Network

Pattern recognition involving large-scale associative memory applications, generally constitutes tightly coupled algorithms and requires substantial computational resources. Thus these schemes do not work well on large coarse grained systems such as computational grids and are invariably unsuited for fine grained environments such as wireless sensor networks (WSN). Distributed Hierarchical Graph Neuron (DHGN) is a single-cycle pattern recognising algorithm, which can be implemented from coarse to fine grained computational networks. In this paper we describe a two-level enhancement to DHGN, which enables it to act as a standard binary image recogniser. This paper demonstrates that our single-cycle learning approach can be successfully applied to denser patterns, such as black and white images. Additionally we are able to load-balance the pattern recognition processes, irrespective of the granularity of the underlying computational network.

Distributed network file storage for a serverless (P2P) network

The design of a persistent peer-to-peer (P2P) network storage application is presented in the paper. SNS separates the application from the underlying P2P network, i.e. the serverless layer. The serverless layer is responsible for routine networking tasks such as self-organisation and maintaining the network state information. The application specific functions are implemented through the file information protocol e.g. maintaining and synchronizing file and disk space information by using lightweight XML-formatted messages. The distributed storage is made to appear as a single large network drive, with all the main features of a conventional file system being made available to the client nodes.

Holographic Graph Neuron: A Bioinspired Architecture for Pattern Processing

In this paper, we propose a new approach to implementing hierarchical graph neuron (HGN), an architecture for memorizing patterns of generic sensor stimuli, through the use of vector symbolic architectures. The adoption of a vector symbolic representation ensures a single-layer design while retaining the existing performance characteristics of HGN. This approach significantly improves the noise resistance of the HGN architecture, and enables a linear (with respect to the number of stored entries) time search for an arbitrary subpattern.

Collaborative-Comparison Learning for Complex Event Detection Using Distributed Hierarchical Graph Neuron (DHGN) Approach in Wireless Sensor Network

Research trends in existing event detection schemes using Wireless Sensor Network (WSN) have mainly focused on routing and localisation of nodes for optimum coordination when retrieving sensory information. Efforts have also been put in place to create schemes that are able to provide learning mechanisms for event detection using classification or clustering approaches. These schemes entail substantial communication and computational overheads owing to the event-oblivious nature of data transmissions. In this paper, we present an event detection scheme that has the ability to distribute detection processes over the resource-constrained wireless sensor nodes and is suitable for events with spatio-temporal characteristics. We adopt a pattern recognition algorithm known as Distributed Hierarchical Graph Neuron (DHGN) with collaborative-comparison learning for detecting critical events in WSN. The scheme demonstrates good accuracy for binary classification and offers low-complexity and high-scalability in terms of its processing requirements.