@umu.ac.in
Associate Professor, Faculty of Computing and Information Technology
Usha Martin University, Ranchi, Jharkhand
Dr. Mohammad Amir Khusru Akhtar is an Associate Professor in the Faculty of Computing and Information Technology, Usha Martin University, Ranchi, Jharkhand, India. His research interest includes mobile ad-hoc network, natural language processing, deep learning and feature engineering. He is the author of over 20 peer-reviewed publications. He received his Ph.D from the Department of Computer Science & Engineering at Birla Institute of Technology, Mesra, Ranchi, India in 2015. He received his M. Tech degree from the Department of Computer Science & Engineering at Birla Institute of Technology, Mesra, Ranchi, India in 2009.
Wireless Sensor Network, IOT, MANET
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Rajesh Kumar Chatterjee, Md. Amir Khusru Akhtar, and Dinesh K. Pradhan
Springer Nature Switzerland
Ajay Kumar, Md. Amir Khusru Akhtar, and Abhishek Pandey
Informa UK Limited
Ajay Kumar, Mohammad Amir Khusru Akhtar, Abhishek Pandey, and Ravi Prakash Srivastava
Informa UK Limited
Mohammad Amir Khusru Akhtar and Gadadhar Sahoo
Elsevier BV
Abstract This paper presents the use of Compressive Sensing (CS) in the reduction of resource consumption to minimize battery and bandwidth usage. It also focuses on how attacks and misbehavior can be nullified. The proposed Neighborhood Compressive Sensing (NCS) model compresses the neighborhood sparse data such as routing table updates, advertisement and trust information. It minimizes resource consumption because major computations are performed by the leader node. The use of compressive sensing gives the reduction in resource consumption because it reduces the amount of transmitting data in the network. It also prevents a network from unwanted advertisement and attacks because the neighborhood nodes do not accept the advertisements and updates directly, rather it uses leader node’s processed information. The proposed NCS model is implemented in “GloMoSim” on top of the DSR protocol, resulting its effectiveness, as compared to the DSR protocol when the network is misconducting for its selfish needs. Simulation result shows that the proposed NCS model is outperformed DSR in terms of the energy consumption, network lifetime and packet dropping ratio. This work is the extended version of Reduction in Resource Consumption to enhance cooperation in MANET using Compressive Sensing (Akhtar and Sahoo, 2015) [68] .
Supriya Gupta and Md. Amir Khusru Akhtar
Springer Singapore
Md. Amir Khusru Akhtar, Mohit Kumar, and Ashwani Kumar
Springer Singapore
Arvind Kumar Sinha, Md. Amir Khusru Akhtar, and Ashwani Kumar
Springer Singapore
Mohit Kumar, Dinesh Kumar, and Md Amir Khusru Akhtar
IGI Global
The prevalent applications of WSN have fascinated a plethora of research efforts. Sensor nodes have serious limitations such as battery lifetime, memory constraints, and computational capabilities. Clustering is an important method for maximizing the network lifetime. In clustering, a network is divided into virtual groups, and CHs send their data to the BS either directly or using multi-hop routing. CHs are some special nodes having more energy than normal nodes. In fact, these special nodes are also battery operated and consequently power constrained; thus, they play a vital role in network lifetime. Cluster formation is very important and improper design may cause overload. This paper presents a modified GA-based load balanced clustering (MGALBC) algorithm for WSN. It is better than GA-based load balanced clustering (GALBC) algorithm because it balances the load by considering the residual energy. The result shows that the proposed method is better than GALBC in terms of energy consumption, number of active sensor nodes and network life.
Mohit Kumar, Dinesh Kumar, and Md. Amir Khusru Akhtar
Springer Singapore
Sink node mobility is to uniformly distribute the load of the sink neighbors in order to balance the energy level and to delegate sink’s neighbor responsibility between the sensor nodes. In this paper, we present a mathematical model that works with existing routing protocol. Sink mobility is decided on the basis of definite-stay-value (\\( {\\text{D}} = \\Delta *(1 - {\\text{P}}(S|Pos)) \\)), where \\( {\\text{P}}\\left( {{\\text{S}}|{\\text{Pos}}} \\right) \\) is defined as closeness of sink node and sink site. Closeness means surrounding density or an area where density of sensors is very high. This model performs the SINK_TEST for the determination of sink site. We have formulated this problem with the help of prior probability and continuous Bayes’ theorem. This mathematical model is verified by experiments and gives adequate accuracy for the improvement of network lifetime in independent environment.
Amir Khusru Akhtar, Gadadhar Sahoo, and Mohit Kumar
IEEE
In corpus preparation we do part-of-speech (POS) tagging where we add POS information into the corpus in the form of tags. The POS information contains a number of tags such as noun, pronoun, verb, adjective, adverbs, preposition, conjunction etc. Literature shows the lack of corpora for Santali language. In this paper we have created and described a Santali language corpus using Sketch Engine corpus query tool that is very much useful for linguistic research. This paper shows statistical information about the Santali corpus such as number of words, tokens and sentences. We have shown lexicons such as number of words, tags, lemmas and lempos. This paper also shows parts of speech tags and lemmatized corpus in terms of noun, numeral, preposition, pronoun, verb, adjective, adverb, conjunction. We have added 590,314 tokens, 425,238 words and 63,199 sentences in our Santali corpus.
Md. Amir Khusru Akhtar, Mohit Kumar, and Gadadhar Sahoo
IEEE
Santali is a language in the Munda or Kolarlan family and has reached a higher stage of development than other language of its family. It is being written in Roman script from a long time and this script has been reoriented to express the peculiar pronunciation and phonetics of Santali language. Automata can be used at all the stages of natural language processing. It has the ability to efficiently characterize morphological and phonological rules. In order to develop a dependency parser for Santali language, we have used automata for morphological analysis. This paper presents the morphology of Santali language. We have used finite-state transducers (FST) for finite-state morphology and discussed initial morphological analysis of nouns and verbs. This paper discusses Santali nouns in terms of gender, number and case. We have shown Verbs in Santali inflect for tense/aspect/mood (TAM), voice and the person and number of the subject. The automaton for the subject and the object markers in first, second and third person have been shown.
Md. Amir Khusru Akhtar and G. Sahoo
Springer India
Energy and bandwidth are the scarce resource in a wireless network. In order to prolong its life nodes drop packets of others to save these resources. These resources are the major cause of selfish misbehavior or noncooperation. To enforce nodes cooperation this paper presents the reduction in resource consumption using Compressive Sensing. Our model compresses the neighborhood sparse data such as routing table updates and other advertisement. We have divided a MANET in terms of the neighborhood called neighborhood group (NG). Sparse data are compressed by neighborhood node and then forwarded to the leader node. The leader node joins all neighborhood data to reconstruct the original data and then broadcasts in its neighborhood. This gives a reduction in resource consumption because major computations are performed at leader end which saves battery power of neighborhood nodes. It compresses sparse data before transmission thus reduces the amount of transmitting data in the network which saves the total energy consumption to prolong life of the network. It also prevents from several attacks because individual nodes do not accept the advertisement and updates directly rather it uses leader node processed information.
Md. Amir Khusru Akhtar and G. Sahoo
Elsevier BV
Abstract MANET is a cooperative network in which every node is responsible for routing and forwarding as a result consumes more battery power and bandwidth. In order to save itself in terms of battery power and bandwidth noncooperation is genuine. Cooperation can be enhanced on the basis of reduction in resource consumption by involving a limited number of nodes in routing activities rather than all. To get accurate selection of nodes to define a backbone several works have been proposed in the literature. These works define a backbone with impractical assumptions that is not feasible for MANET. In this paper we have presented the Backbone Group (BG) model, which involve the minimum number of nodes called BG in routing activities instead of all. A BG is a minimal set of nodes that efficiently connects the network. We have divided a MANET in terms of the single hop neighborhood called locality group (LG). In a LG we have a cluster head (CH), a set of regular nodes (RNs) and one or more border nodes (BNs). The CHs are responsible for the creation and management of LG and BG. The CHs use a BG for a threshold time then switches to another BG, to involve all nodes in network participation. The proposed model shows its effectiveness in terms of reduction in routing overhead up to a ratio (n2: n2/k) where k is the number of LGs.
Md. Amir Khusru Akhtar and G. Sahoo
Springer New York
MANETs may be considered as a society in which nodes agree to cooperate with each other to fulfill the common goal. But noncooperation is genuine to save itself in terms of their battery power and bandwidth. Ad hoc network is still a challenge as lots of work has been proposed but they have serious limitations in terms of routing overhead and attacks. Modification of routing information’s can be handled by secure routing protocols but non cooperation is still in its natal stage. Our proposed FG Model minimizes resource utilization by cutting down the routing overhead, so that less number of nodes participates in routing activities by creating smaller friendly groups. The proposed method minimizes routing overhead to a ratio of \\(k{:}\\;1|k>1\\) (where k is the no. of friendly groups) and it can be implemented on the existing MNAET routing protocol (e.g., DSR, AODV and TORA).