Liren Zhang

Characteristics of network delay and delay jitter and its effect on voice over IP (VoIP)

Voice over IP is a hot topic for the networking technologies. Delay and delay jitter are the important measures of QoS particularly for the voice traffic in the network environment with bursty background traffic. This paper studies the performance behavior of delay and delay jitter at the IP packet level by using the transient queuing solution. The analysis focuses on the voice over IP traffic going through a IP router with the bursty background traffic over network. We study the individual effects of various traffic parameters on the jitter behavior of packet voice stream multiplexing background traffic with different bursty characteristics in the IP router. The numerical results for the investigation of the effects of delay and delay jitter to a speech sample are presented.

Effect of delay and delay jitter on voice/video over IP

Internet, especially in the situation when network has bursty background traffic. However, it is usually difficult to evaluate the statistics of delay and delay jitter in such bursty traffic environment due to their transient queuing nature. This paper focuses on the performance behavior of delay and delay jitter at IP packet level and also their effects on the performance of voice and video packet streams when they are multiplexed with bursty background traffic in IP routers. The performance analysis of delay and delay jitter is done using the transient queuing solution technique. The numerical results are presented in terms of probability density functions (pdfs) of delay and delay jitter, which focus on their effects on speech samples and video frames.

A Novel Location-Service Protocol Based on

This paper presents a novel k-hop cluster-based location service (KCLS) protocol in mobile ad hoc networks, which is able to well balance the tradeoff between the communication overheads and the accuracy of location information. Other advantages of the proposed KCLS protocol include excellent network scalability and tolerance of link breakage that can significantly reduce communication overheads and the latency caused by route path recovery. The numerical results obtained from both theoretical analysis and simulations have shown that the cost of location management using the KCLS protocol is less than 2% of the location-management cost using the link-state protocol. Especially for a large value of k, the proposed KCLS protocol is not only able to suppress the increasing rate of the total cost when the number of hosts in the network increases but also increases the hit probability of location service and reduces the passive effect of host mobility on control overhead as well. With good scalability and the capability of self-discovery, the proposed KCLS protocol is adaptable to accommodate most applications in ad hoc networks

k

This paper focuses on the optimization of network bandwidth allocation and buffer dimensioning to transport pre-stored MPEG video data from source to playback destination across ATM networks. This is one of the most important issues in the support of video-on-demand (VoD) service. This paper provides a novel scheme in the dynamic allocation of bandwidth to segments of video using ABR mode. The dynamic bandwidth allocation is based on a new concept, called playback tunnel which is obtained from the traffic characteristics of the pre-stored MPEG video trace to determine the optimum of transmission bandwidth as well as the buffer capacity to ensure that the playback buffer neither underflows nor overflows. The proposed scheme is tested with real-life MPEG video traces. The obtained results have shown its significant performance improvement in terms of the capacity of playback buffer, the start-up playback delay, the size of video segment and the network multiplexing gain.

-Hop Clustering for Mobile Ad Hoc Networks

In this paper, we systematically present the methodology of modeling bursty traffic sources using the 2-state Markov Modulated Bernoulli Process (called MMBP-2). The technique used can be easily extended to an m-state MMBP though the numerical calculation is complicated. We first defined the parameters of the model and some processes associated with it, and subsequently examined the queue length distribution of an infinite buffer driven by an MMBP-2 with batch arrivals. We next looked at the case where the same buffer is fed by a group of two identical MMBP-2 sources. Instead of deriving the queue length expression, we cast the problem in the framework of the previous case and made use of the previous results with some modification. Lastly, we looked at the case of a finite buffer driven by two MMBP-2 sources with different parameters. We formulated the queue length solution in the framework of Markov theory and calculated the Cell Loss Probability (CLP) for this case.

A novel scheme of transporting pre-stored MPEG video to support video-on-demand (VoD) services

Two novel MAC schemes are proposed to solve the hidden terminal problem and the exposed problem in ad hoc networks with large interference range. With the aid of dual busy tones, one scheme is a simple enhancement to IEEE 802.11, and the other is to exactly signify the interference range by adjusting the transmission power of busy tones. Both of the two schemes are able to prevent the collisions of data/ACK packets. The numerical results show that the two proposed schemes are able to enhance the saturation throughput of IEEE 802.11 significantly. The proposed VPDBT MAC can balance the trade off between spatial reuse and collision avoidance.

Research note: Methodology for traffic modeling using two-state Markov-modulated Bernoulli process

This paper focuses on the upper bound of broadcast coverage area to achieve high delivery rate and low number of rebroadcasts. In order to improve the performance of geometry-based protocols, a novel relative position-based broadcast relay (RPBR) approach is proposed, in which the forwarding hosts are able to rebroadcast packets effectively, especially for heavy host density. The numerical results show that the proposed RPBR scheme is able to enhance the distance-based protocol to rebroadcast packets effectively by increasing the packet delivery ratio up to 10%. On the other hand, for the angle-based protocol associated with RPBR, the number of forwarding hosts can be reduced by 30%. A performance comparison of the topology-based broadcast protocols and the RPBR scheme combined with geometry-based in dynamic ad hoc network environment is presented.

IEEE 802.11 MAC protocol enhanced by busy tones

With the introduction of Generalized multiprotocol label switching in the common control plane for optical crossconnects and electrical switching devices, the integration of IP and dense wavelength division multiplexing can be efficiently achieved. When IP traffic flows are directly mapped to optical channels without SDH inter-layer, the bursty nature of IP traffic may have significant effect on the performance of the optical channel. On the other hand, the nonlinear effects of the optical channel may also have significant effect on the performance at IP level. This paper focuses on the effect of fiber nonlinearities on the performance of IP traffic. Numerical results including IP packet error probability and high-order distribution functions of IP packet error are presented in terms of IP traffic load, input light signal power and the frequency space between the optical wavelengths. It is demonstrated that when IP traffic load is light in such systems, the effect of either four-wave mixing (FWM) or stimulated Raman scattering (SRS) is much less serious than the worst-case assumption, and the limitation on allowable power of input light is also relaxed. The obtained numerical results demonstrated that FWM and SRS are both sensitive to power level of input light and frequency spacing, but each in a unique way. Effort must be taken in system design to avoid improving the performance of one effect at the expense of deteriorating the other.

An efficient broadcast relay scheme for MANETs

When setting up a sensor network, one of the first requirements is to establish cryptographic keys for later use. However, the traditional key establishment techniques cannot be directly applied due to the inherent properties of sensor networks. Recently, a promising methodology, random key pre-distribution schemes based on symmetric cryptography, has been proposed. In this paper, the authors studied the problem of replication attack on random key pre-distribution schemes. Using a combination of modeling, analysis, and experiments, the relationship among the replicated hostile nodes, the sensor networks, and the resiliency of various random key pre-distribution schemes were analyzed, characterized, and discussed against replication attack. Example findings include: (1) the sensor networks with random key pre-distribution schemes, even with one replicated sensor, start to become almost 100% insecure when the adversary captures and stores the key information equivalent to those carried by one good sensor node; and (2) among the proposed schemes, the q-composite scheme with larger q is most resilient against replication attack while the basic scheme is least resilient and the Blom-based scheme lies in between the above two schemes when the replicated node has less memory to store key information than the original node. Interestingly, it is the other way round when the replicated node has more memory to store key information than the original node. Moreover, as a transition, the resilience against replication attack is the same for all the random schemes when the replicated and original nodes have the same memory to store key information. This study does not only provide practical insights into the design of more secure and efficient key establishment schemes allowing simple key establishment for large-scale sensor networks but can also be used to accurately predict the payoff that an adversary can gain after injecting a certain number of nodes into the sensor networks.

Optical nonlinear effects on the performance of IP traffic over GMPLS-based DWDM networks

This paper focuses on the design of a novel approach for dynamic k-hop clustering architecture, which is called k-hop compound metric based clustering (KCMBC). KCMBC is a fast convergent and load balancing clustering approach that is able to demonstrate significant improvement on the network performance, in terms of scalability and stability for large-scale ad hoc networks. On the other hand, since KCMBC has taken into account the host mobility, the clusters constructed by KCMBC are more stable than the other schemes. Simulation results show that the clusters created by using the KCMBC approach have modest but more uniform cluster size. Moreover, cluster-head duration can be increased by KCMBC significantly.