Muhammad Usman Ilyas

A distributed and privacy preserving algorithm for identifying information hubs in social networks

This paper addresses the problem of identifying the top-k information hubs in a social network. Identifying top-k information hubs is crucial for many applications such as advertising in social networks where advertisers are interested in identifying hubs to whom free samples can be given. Existing solutions are centralized and require time stamped information about pair-wise user interactions and can only be used by social network owners as only they have access to such data. Existing distributed and privacy preserving algorithms suffer from poor accuracy. In this paper, we propose a new algorithm to identify information hubs that preserves user privacy. The intuition is that highly connected users tend to have more interactions with their neighbors than less connected users. Our method can identify hubs without requiring a central entity to access the complete friendship graph. We achieve this by fully distributing the computation using the Kempe-McSherry algorithm to address user privacy concerns. To the best of our knowledge, the proposed algorithm represents an arguably first attempt that (1) uses friendship graphs (instead of interaction graphs), (2) employs a truly distributed method over friendship graphs, and (3) maintains user privacy by not requiring them to disclose their friend associations and interactions, for identifying information hubs in social networks. We evaluate the effectiveness of our proposed technique using a real-world Facebook data set containing about 3.1 million users and more than 23 million friendship links. The results of our experiments show that our algorithm is 50% more accurate than existing distributed algorithms. Results also show that the proposed algorithm can estimate the rank of the top-k information hubs users more accurately than existing approaches.

Measurement Based Analysis and Modeling of the Error Process in IEEE 802.15.4 LR-WPANs

Knowledge of the error process and related channel parameters in wireless networks is invaluable and highly instrumental in a broad range of applications. Under the IEEE 802.15.4 Low Rate-Wireless Personal Area Networks (LR- WPAN) standard, compliant devices are capable of providing two pieces of information about the channel conditions along with each received packet, the Link Quality Indication (LQI) and Received Signal Strength Indication (RSSI). Together they constitute a form of Channel State Information (CSI). This work is based on statistical and information theoretic analysis of a very extensive data set of wireless channel traffic between a transmitter and receiver, called packet traces. Data is collected in a variety of documented environments. To our knowledge, this is the first detailed trace collection effort for this type of network. The traces distinguish themselves from data sets of other studies in that they record individual bit errors as well as packets that are never detected by receivers. First, we provide a detailed analysis of the IEEE 802.15.4 wireless channel. More specifically, we provide a detailed analysis of the Bit Error Rate (BER) process at individual bit and on a packet-by-packet basis. We explore the relationship between the packet-level BER process and the LQI and RSSI processes (also observable on a packet-by- packet). The analysis shows that measurements of both LQI and RSSI provide information that allows us to reduce uncertainty about the BER. Secondly, we develop a model of the BER process that is driven by observable CSI parameters. Thirdly, we continue our analysis with measurements of channel memory at the packet and bit level. We determine that the wireless channel 2 bits memory. At the packet level we observe that the amount of channel memory is more varied.

Identifying Leaders and Followers in Online Social Networks

Identifying leaders and followers in online social networks is important for various applications in many domains such as advertisement, community health campaigns, administrative science, and even politics. In this paper, we study the problem of identifying leaders and followers in online social networks using user interaction information. We propose a new model, called the Longitudinal User Centered Influence (LUCI) model, that takes as input user interaction information and clusters users into four categories: introvert leaders, extrovert leaders, followers, and neutrals. To validate our model, we first apply it to a data set collected from an online social network called Everything2. Our experimental results show that our LUCI model achieves an average classification accuracy of up to 90.3% in classifying users as leaders and followers, where the ground truth is based on the labeled roles of users. Second, we apply our LUCI model on a data set collected from Facebook consisting of interactions among more than 3 million users over the duration of one year. However, we do not have ground truth data for Facebook users. Therefore, we analyze several important topological properties of the friendship graph for different user categories. Our experimental results show that different user categories exhibit different topological characteristics in the friendship graph and these observed characteristics are in accordance with the expected ones based on the general definition of the four roles.

Header Detection to Improve Multimedia Quality Over Wireless Networks

Wireless multimedia studies have revealed that forward error correction (FEC) on corrupted packets yields better bandwidth utilization and lower delay than retransmissions. To facilitate FEC-based recovery, corrupted packets should not be dropped so that maximum number of packets is relayed to a wireless receivers FEC decoder. Previous studies proposed to mitigate wireless packet drops by a partial checksum that ignored payload errors. Such schemes require modifications to both transmitters and receivers, and incur packet-losses due to header errors. In this paper, we introduce a receiver-based scheme which uses the history of active multimedia sessions to detect transmitted values of corrupted packet headers, thereby improving wireless multimedia throughput. Header detection is posed as the decision-theoretic problem of multihypothesis detection of known parameters in noise. Performance of the proposed scheme is evaluated using trace-driven video simulations on an 802.11b local area network. We show that header detection with application layer FEC provides significant throughput and video quality improvements over the conventional UDP/IP/802.11 protocol stack

Increasing Network Lifetime Of An IEEE 802.15.4 Wireless Sensor Network By Energy Efficient Routing

In a multi-hop 802.15.4 Wireless Sensor Network the volume of traffic that is processed by intermediate nodes increases considerably as data approaches the PAN coordinator. This problem is referred to as the reachback problem. In this work we focus on a more fundamental implication of reachback on operational network lifetime of 802.15.4-based wireless sensor networks. This paper proposes the joint minimization of two network lifetime metrics, previously proposed for independent use, to more accurately characterize the degree of longevity of wireless sensor networks. The method uses the k-shortest simple path algorithm and a dynamic programming method rooted in operational rate-distortion (RD) theory to increase the operational lifetime of wireless sensor networks.

A KLT-inspired node centrality for identifying influential neighborhoods in graphs

We present principal component centrality (PCC) as a measure of centrality that is more general and encompasses eigenvector centrality (EVC). We explain some of the difficulties in applying EVC to graphs and networks that contain more than just one neighborhood of nodes with high influence. We demonstrate the shortcomings of traditional EVC and contrast it against PCC. PCCs ranking procedure is based on spectral analysis of the networks graph adjacency matrix and identification of its most significant eigenvectors.

A Distributed Algorithm for Identifying Information Hubs in Social Networks

This paper addresses the problem of identifying the top-k information hubs in a social network. Identifying top-k information hubs is crucial for many applications such as advertising in social networks where advertisers are interested in identifying hubs to whom free samples can be given. Existing solutions are centralized and require time stamped information about pair-wise user interactions and can only be used by social network owners as only they have access to such data. Existing distributed algorithms suffer from poor accuracy. In this paper, we propose a new algorithm to identify information hubs that preserves user privacy. Our method can identify hubs without requiring a central entity to access the complete friendship graph. We achieve this by fully distributing the computation using the Kempe-McSherry algorithm, while addressing user privacy concerns. We evaluate the effectiveness of our proposed technique using three real-world data set; The first two are Facebook data sets containing about 6 million users and more than 40 million friendship links. The third data set is from Twitter and comprises of a little over 2 million users. The results of our analysis show that our algorithm is up to 50% more accurate than existing algorithms. Results also show that the proposed algorithm can estimate the rank of the top-k information hubs users more accurately than existing approaches.

On enabling cooperative communication and diversity combination in IEEE 802.15.4 wireless networks using off-the-shelf sensor motes

This paper presents the ‘Generalized Poor Man’s SIMO System’ (gPMSS) which combines two approaches, cooperative communication and diversity combination, to reduce packet losses over links in wireless sensor networks. The proposed gPMSS is distinct from previous cooperative communication architectures in wireless sensor networks which rely on a relay channel, and also distinct from implementations in 802.11 networks that require a wired infrastructure or hardware changes for cooperation. gPMSS foregoes the need for any changes to mote hardware and it works within the current IEEE 802.15.4 standard. We describe the gPMSS protocol that governs the cooperation between receivers. Three variants are evaluated including selection diversity, equal gain and maximal ratio combining. First, we demonstrate gPMSS on bit error traces in a fully reproducible manner. This is followed by an implementation of gPMSS in C# on the .NET Micro Framework edition of the recently released Imote2 mote platform. We demonstrate by means of experiments an increase in the packet reception rate from 22–30% to 73–76%, a relative increase of 150–245%. We also analyzed the power consumed by the transmitter per delivered packet and observe a reduction of up to 68%. We also take into account the retry limit of the IEEE 802.15.4 protocol and demonstrate that gPMSS is able to provide 99% packet delivery at the protocol’s default retry parameters against 65–75% without it.

Long Range Dependence of IEEE 802.15.4 Wireless Channels

Several works have attempted to determine whether the process that introduces errors into transmissions over a wireless channels is long range dependent (LRD). Most of these efforts were focused on IEEE 802.11b/g channels. Knowledge of LRD is of significant interest to designers of wireless network systems. This experiment-based study attempts to gain a better understanding of the standardized IEEE 802.15.4 low rate-wireless personal area network (LR-WPAN) specification. We determine whether the bit, symbol and packet level error processes are LRD or not. Is the memory length of the channel measurable and finite, or is it sufficiently long to qualify it as an LRD process? We record channel behavior by capturing all transmissions at the receiver including the ones that would under normal operation be dropped and use these to generate residual error traces. We perform statistical data analysis of these traces using various tools for scenarios with and without interference from 802.11b/g sources. For the bit and symbol level error process we conclude that channel memory remains fixed at 2 bits and 2 symbols respectively regardless of channel and environment. For the packet level error process we conclude that the perception of on and off LRD is actually due to interference from co-located 802.11b/g networks.

Who are you talking to? Breaching privacy in encrypted IM networks

We present a novel attack on relayed instant messaging (IM) traffic that allows an attacker to infer whos talking to whom with high accuracy. This attack only requires collection of packet header traces between users and IM servers for a short time period, where each packet in the trace goes from a user to an IM server or vice-versa. The specific goal of the attack is to accurately identify a candidate set of top-k users with whom a given user possibly talked to, while using only the information available in packet header traces (packet payloads cannot be used because they are mostly encrypted). Towards this end, we propose a wavelet-based scheme, called COmmunication Link De-anonymization (COLD), and evaluate its effectiveness using a real-world Yahoo! Messenger data set. The results of our experiments show that COLD achieves a hit rate of more than 90% for a candidate set size of 10. For slightly larger candidate set size of 20, COLD achieves almost 100% hit rate. In contrast, a baseline method using time series correlation could only achieve less than 5% hit rate for similar candidate set sizes.