QOS Protocols For Estimation Of Resources Computer Science

Essay add: 19-06-2017, 11:42   /   Views: 9

An IEEE 802.11 networks has been proved to provide a certain level of quality of services by the means of service differentiation, but due to the changes and amendments in the IEEE 802.11 e. No method or accurate mechanism has been proven to provide the standard and accurate levels of available amount of resources in the network, even some irresponsive fields in the network may effect the estimations. Such an estimation level of available amount of resources in the network may be a good strength to the bandwidth-constrained applications. Due to the irrespective of structure and mobility of the nodes in multi hop ad hoc networks this estimations become even more difficult. The estimation of available bandwidth in accuracy remains interesting. In this project the author proposes a new mechanism to estimate the accurate bandwidth estimation in the ad hoc networks and to find out the factors influencing the network resources and this mechanism has been shown through the java simulations.


Ad hoc networks are self organised, self autonomous, wireless and mobile networks. They do not require any fixed access points as like with the wired networks, as the nodes organizes themselves and automatically manage themselves to transfer the data between them and they automatically manage the topologies due to the mobility nature in them. Many of the current and mobile ad hoc networks assume underlying the IEEE 802.11 standard because of having broad availability of interface and simulation models between them. As the transmission range is limited by the regulations, a distributed routing protocol is required to extend or increase the range of the data transmissions in the networks. These standards have not perfectly suited for the multi-hop ad hoc networks. Nowadays, several application constraints produce multimedia data flows or rely on the proper and efficient data transmission of sensitive control traffic. These application constraints may benefit from the (QoS) quality of service support in the network. QoS gathers several concepts and theories to define a fine quality of protocols to run the network. Some protocols offers very strong guarantees to the applications on the data transmission characteristics, for example rely of the packets, bandwidth, or network load. Other solutions to the ad hoc networks work only the selection of the best route available to transmit the packet. In these both cases an accurate estimation of band width is required. Most of the applications related to the QoS level of services leaves this problem and concentrate on the network load and the rely of the packets and leaving this problem to the link layer and are assumed the link layer can estimate the bandwidth of the network to the accurate level but they are not in the reality. The accuracy of the bandwidth of the network assumed by the link layer is not much accuracy in the real time. This resource availability estimation problem is far from being trivial as it must take the wireless environment into account and several phenomena related to it and but also dependent on the less amount of the parameters like the node mobility and the structure of the network.

Throughout this research the author focuses on the fundamental and basic resources available: throughput and estimating the bandwidth of the network. But this is quit tricky because of the wireless network, because the medium is shared among all the nodes. Computing the bandwidth estimation in the network and between two neighbour nodes needs the accurate identification of all potential participators from the emitter's side, of all the potential scramblers on the receiver's side and a proper estimation of their impact. Information about the utilized nodes in the shared resource is gathered and so the nodes which are free in the network can be derived. Both the tasks are too difficult because in the ad hoc networks two nodes may share the medium at a time even they may not transmit the data directly. In this research the author proposes a new mechanism of bandwidth estimation in the IEEE 802.11 MAC layer ad hoc networks. This mechanism uses the carrier sense capability of the nodes in the resource and combined with collision avoidance and providence mechanism as well. This generates the upper layers as an evaluation that comprises of both the accuracy of the bandwidth and measurement cost. And this application is highly adaptable with some other medium access protocols. In the wireless ad hoc networks the new flow may possibly take over the capacity of the entire network due to the strong mutuality between the wireless links. Therefore, the collisions may occur while the evaluation started so it is mandatory to take the disruptions into the account and to differentiate between the raw throughputs that may transferred to the maximum data throughput that may transferred without any noticeable interference. So, the available bandwidth between two peers is defined as the maximum throughput between those peers without interrupting the any ongoing data flows in the network. Link capacity is denoted by the maximum throughput a flow can achieve between two peers regardless of other flows present in the network.


First the investigation has to be done on the QoS protocols for estimation of resources available on the network and then a research has to be done on the irresponsive applications which are disrupting the estimation and evaluating the bandwidth of the network. Then a new mechanism should be proposed to estimate the resource available in the network with more accuracy than the previous existed protocols.


Through the java simulations the estimated bandwidth is to be evaluated according to the new proposed mechanism.

LITERATURE REVIEWANALYSIS OF REFERENCESTheoretical concepts of wireless adhoc networks (Shih-lin-wu, yu-chee-tseng, 2007)

Form the above I have gone through all the basic concepts of IEEE 802.11 based ad hoc networks. The book clearly explains about the autonomous features of the wireless ad hoc networks. This book mainly helped me on the key concepts of wireless ad hoc network.

Routing protocols for ad hoc networks (George Aggelou, 2005)

From the above book helped me in learning the routing protocols of wireless ad hoc networks. In this book I have keenly learned about the route discovery, route maintenance and routing protocols.

Protocol and system measurement (chadi Barakat, Ian Pratt, 2004)

The above book helped in studying the concepts behind the measurements of the basic resources available in the wireless ad hoc networks. The topics like origination of Micro congestion at the Access point helps in this domain of research.

Active bandwidth estimation techniques (R. Prasad, M. Murray, C. Dovrolis, and K. Claffy, 2003)

This is an IEEE transaction discussion helped to me to study about the estimation techniques used in the wireless networks and mainly this journal helped me in studying on the active bandwidth estimation techniques

Different protocols for estimation os resource availability (M. Jain and C. Dovrolis, 2004)

This IEEE discussion gave the clear idea about the existed protocols for which they were used to measure the available bandwidth of the network and in this the book mainly explained about the SLoPS protocols

One of the technique for measurement (B. Melander, M. Bjorkman, and P. Gunningberg,2000)

This resource gave the another technique to estimate the bandwidth available and in this I have clearly understood the concept of TOPP protocol for measuring the resources available.

Congestion measurement and estimating (F.Y. Li, M. Haugea, A. Hafslund, O. Kure, and P. Spilling, 2004)

In this resource I studied about the detection of congestion by the probe packets technique to see the exact delay in the network and if it crosses to the paper estimated delay therefore a congestion is occurred in the network.

Other impacts on bandwidth estimation (A. Johnsson, B. Melander, and M. Bjo¨rkman, 2005)

The above resource helped me in finding the other factors which made difficult in estimation of bandwidth for wireless ad hoc networks. The volume of cross-traffic flow and like factors influences the bandwidth estimation which is very negligible in the wired network than the wireless networks.


Available bandwidth evaluation has generated several concepts in the wireless network technologies and the wired technologies as well. There are several classifications proposed and they are put forward as some different categories.


Active approach techniques that rely on the emission of dedicated end-to-end packets to estimate the available bandwidth along the path between the nodes in the network.


These techniques use the local information to estimate the bandwidth available. Normally these are transparent and uses exchange information between the nodes via one-hop broadcasting technique.

The following are the existing protocols and techniques used to estimate and evaluate the bandwidth available in the network.


Mainly these techniques measure the end-to-end available bandwidth by sending packets of equal size from source to the destination. The source slowly increases the probe packets emission rate. Measurements of this performance and characteristics has been done on the receiver's side and then converted to the end-to-end bandwidth estimation. There are some other protocols falls into this category like SLOPS AND TOPP. These protocols are mainly differing in the way of increasing the probe ratio. These techniques are worth failing because the probe packet flow may influence the existing flows. Li et al propose to detect the presence of congestion by monitoring the delay of the probe packets. Whenever this delay increases than the theoretical delay then the medium suffers from the congestion. Then depending upon these measurements, they computed the medium utilization and then derived the channel capacity and channel usage ratio. The TOPP protocol model clearly explains that the probe packet size and the cross-traffic in the network impact the measurement of the bandwidth more in the wireless networks than the wired networks. So, therefore these technologies are very sensitive in measuring the network parameters in the wireless ad hoc networks.


There is a technique called dynamic bandwidth management scheme for single-hop networks. In this one of the nodes in the network hosts the bandwidth manager process and this node is the responsible for calculating the bandwidth estimation in the cell and allocates the bandwidth to each peer in the network. By using the control messages each peer requests the bandwidth manager for an exclusive usage of the channel during a proportion of time. As the structure of the network is reduced to the single-channel, the available time-share is calculated by the total loads i.e., the sum of all individual loads in the network. Then the available fraction of time is translated into an available bandwidth by considering the capacity of the wireless link called the total bandwidth of the network.

This approach can be considered as passive because very few data packets are participated. This solution has been granted to evaluate only to the single hop networks but not the multi-hop and ad-hoc networks.


This is also a different technique proposes an available bandwidth estimation on per-node performance. The evaluation mechanism constantly updates a value called bandwidth efficiency ratio (BWER). Bandwidth efficiency ratio is the ratio between the number of packets transmitted and received. Then the available bandwidth derived by multiplying the ratio with channel capacity. This ratio value is broadcasted among all the single hop neighbors through hello messages. The available bandwidth to the node is the assigned as the minimum of the available bandwidth over a closed single-hop neighborhood. QoS-AODV considers not only sending the possibility to send a given amount of data but also considers the effect of the emission of the node on its neighborhood.


Chandet and Lassons proposed a bandwidth reservation protocol called bandwidth reservation under interferences influence (BRuIT). This ABE protocol takes the IEEE 802.11 as a standard; the carrier sense range is larger than the communication range. Emitter shares the medium and bandwidth with other nodes but they cannot participate in the data communication. An experimental study on this issue shows that the carrier sense range is at least twice the communication range radius. Each node regularly broadcasts the bandwidth to all its immediate neighbor nodes to route and to emit flows and estimated bandwidth. It also broadcasts the similar information considering all its one-hop neighbors, propagating such information at a two-hop distance. When the carrier sense radius is exactly equal to the twice of the communication sense radius the authors have shown that best compromise is done between accuracy, estimation and cost of the bandwidth evaluation.


The Contention Aware Admission Control Protocol is also a technique to evaluate the bandwidth available in the network. In this technique each node calculates its local proportion of idle channel time by monitoring the radio medium. Then this protocol propagates three different techniques to propagate the information to the maximum number of nodes within the carrier sense radius. The first is bit same with the BRuIT protocol, propagating the hello messages including the information to reach two-hop neighborhood. Second, they increase the nodes transmission power but this power is limited and regulated by the power control mechanisms and this technique is more applicable when the power control transmission is used for regular transmissions. Finally depending upon the signal modulation and quality of the electronics the receiver's side nodes reduces their sensitivity to decode the information coming from the sender's side. This protocol also includes about the existence of intraflow contention. When a flow takes a multi-ho routes, successive routers contend for the channel access for the frames belonging to the same flow in the network so, it is thus important to take the route length in to the account when performing the admission control protocol.


Adaptive Admission Control Protocol (AAC), in this protocol it makes each node as the set of all contender potential nodes as a single node. It measures the activity time period durations and considers it as a frame emission of the corresponding length. With the help of this mechanism, collisions and distinct emissions are also considered when computing the medium occupancy. Based on this measurement each node is able to evaluate its available bandwidth. It exchanges this information to all the neighbor nodes to compute its bandwidth on each link, here link is defined as the set of two nodes. This value is defined as the minimum bandwidth available between the both ends. AAC also takes the intraflow contention into account as the CACP protocol does.


Based on the drawbacks of the present existing system the following concepts determine the evaluation of the bandwidth in an improved manner.


The carrier sense mechanism: - This mechanism prevents two close emitters from the transmitting simultaneous transmissions, unless they draw the back-off counter value. Therefore, an emitter shares the bandwidth with all its close residing nodes. The channel utilization has to be monitored to evaluate the capacity of a node or peer to emit in a given traffic volume.

For a transmission to take place the receiver needs to contend that no interference should occurs during the whole data transmission. Therefore the value of the bandwidth estimation on a link depends on both peer channel utilization ratio and also depends upon the idle time synchronization. The synchronization needs to be evaluated.

No collision detection is possible in the wireless networking systems. Therefore, whenever a collision occurred both colliding frames are completely emitted and maximizing the bandwidth loss in the system. The collision probability needs to estimated and integrated to the ABE (Available bandwidth Estimation)

When the collision takes place in the single or unicast frames, the IEEE 802.11 protocol automatically retrieves to emit the same frame, drawing the back off counter in a double-sized contention window. The time lost in the additional overhead may also have an impact on the estimating the bandwidth and that influence factors has to be evaluated.



Carrier Sense Mechanism: - Estimating a node's emission capabilities. Whenever a node wants to transfer a data first it needs to contend for the medium access and without having the shared medium the node cannot transfer the data. Therefore, whenever a potential node wants to send a frame it should have some shared medium to do that it has to calculate the load of the medium i.e., the proportion time of the medium is idle to determine the chance of the successful gain access in the shared medium resource. Let us consider a node a in the network during an observation interval of seconds.

Tidle(a) is the total Idle time it is the total time during the node a neither emits the frame nor uses the shared medium ( sensing the medium is busy). This also includes the periods during which no frame is ready to transfer as well as periods of deferral (back off time and interframe spacing).

Bs is the bandwidth the available to node a, i.e., the maximum throughput of the node that can transmit without degrading close flow's rate.

Cmax is the maximum capacity of the medium.

During an arbitrary observation interval, each node may monitor the radio medium in its surroundings and it calculates the total amount of time of idle for emitting the frames and to participate in the data transmission. To adapt the evaluation to the MAC IEEE's behaviour the periods of time should be shorter that IEEE 802.11 DIFS timing and shall not be added to the total idle time count, as such intervals do not allow any back off decrease nor medium access. As soon as the signal receives the threshold of the carrier sense range then the medium is said to be busy. This method not only takes the bandwidth used in the transmission range of the nodes but it also takes the whole carrier sense range area in to the account. As the monitoring neither takes the IEEE 802.11's variable in to the account nor the reception side of the transmission, the available bandwidth computed by this method at node a is inaccurate. However, it provides the threshold value above of the medium access probability decreases rapidly. Some frames are successfully transmitted due to the favourable scheduling of the transmission or due to capture effects. As long as the medium load is below the threshold value, a scheduling between the two different contender nodes or emitters prevents the simultaneous emissions exists. So this value can be considered as the upper bound of the available bandwidth

. Cmax

Idle Period Synchronization: - The above mechanism evaluated the upper bounds of the bandwidth a node could use to emit frames. The reception part of the transmission also requires the medium to be free during the transmission, thus the previous measurement of the threshold should also be considered at the receivers' side also.

Consider a radio link composed of two neighbour nodes a, b. In order to use combinational tools, the mechanism needs to consider all the time is to discrete.

Be the time spacing referring as a time unit.

Tm= is the number of units in a measurement period.

Ta and Tb are the number of units during which the medium is available for nodes a, b.

Ba, Bb are the two true available bandwidth bounds for the nodes a, b and are calculated as per the carrier sense mechanism.

B(a,b) is the true available bandwidth on the link (a, b) i.e., the real bandwidth that can be achieved without degrading close flows.

b(a,b) are the estimated available bandwidth value on the link (a, b).

If Ba is null or close to zero then it means the node a never gains the access to the medium or it might already be emitting the frames at a rate that same of load capacity of the radio medium. Similarly at receiver's side if the medium is busy then the frames systematically experiences the collisions and the communication and data transmission never succeeds or it might be the B(a,b) min(Ba, Bb). If sending a flow with a throughput higher than the min(Ba, Bb) necessarily provokes a medium exact load around a and/or b. Considering this minimum value as the available bandwidth may also lead to an over estimation and evaluation of bandwidth. In general the idle periods at emitters and receivers sides are desynchronized.

In ad hoc networks due to the complex interactions between the nodes, emitters and receivers the time synchronization are unlikely to be perfectly synchronized. And evaluating the impact of this synchronism requires the exchange of the exact medium utilization and a fine clock between the peers and nodes which generates a high of over head. Due to this reason here the author proposes the probabilistic mechanism to estimate the effect of the phenomena.

The requirements for a successful frame transmission are as follows.

First for the communication to start, the medium needs to be free for at least DIFS on the emitter's side so that this emitter gains the access for the medium to share. On the receiver's side the medium has to be free during the time required to transmit the whole data frame (Tdata); otherwise, a collision occurs in between the data transmission and though this value is not perfectly accurate in real time because it accepts that the level of signal that would provoke a collision equal to the carrier sense threshold, regardless of the distance between the emitter and the receiver.

Consider a uniform random distribution of the medium occupancy over the observation period so it then possible to compute the expected delay E (l(r, s)). Nodes s and r sense the medium idle simultaneously. Consider P (i, j, k) as the probability to be estimated.



From the above expression, the probability P (l(r,s)=i) that the first synchronization occurs at a given time unit and the before the synchronization the expected delay will be E (l(r,s)).

P (l(r,s)=i) =

E (l(r, s)) =

Still considering the mediums uniform random distribution of the occupancy, the available expected bandwidth E (l(r, s)) can be evaluated by expressing the probability that the medium is free simultaneously at the emitter's and receiver's sides

P (b(s, r)=i) = ,

E (b(r, s)) =.

ABE: -

The different points discussed above can be combined to estimate the available bandwidth on a wireless link i.e., between a given emitter and a given receiver. The whole mechanism is called ABE and because of its design it is also called as the lightweight protocol. The available bandwidth between two neighbour nodes s and r can be evaluated by the following.

Efinal(b(s, r)) = (1-k) . (1-P) . E (b(s, r)).

Where E(b(s, r)) is the available bandwidth on link(s, r) evaluating by monitoring the radio channel and combining emitter's and receiver's values in a probabilistic manner, P is the collision probability between measured on the received hello packets and rescaled to the appropriate packet size k and k is the proportion bandwidth lost due to the backoff scheme computed due to P.


In the ABE-AODV, neighboring nodes exchange their available bandwidth via hello messages which is computed locally by the servers. For every second, each node locally estimates its medium occupancy ratio and includes this information in a hello packet. And then these values are converted in to the link evaluation using ABE-estimating protocol. The accuracy of the bandwidth depends upon the value of which can be considered as sampling period. By hiding the more and fast variations in the medium, the large amongst of the sampling of more the accuracy can be calculated. However, the sampling unit should be always small enough to allow fast reactions to long term load variation and the long term node mobility's.

However Hello based packet techniques may generates additional overhead depending upon the Hello emission frequency. The Hello packet emission frequency should be adapted node mobility to the flow dynamics. In order to have the meaningful comparisons the author fixed the value to =1 second in ABE-AODV. Similarly everything is tuned accordingly to emit one information frame for each second.


Integration of ABE-AODV

The author slightly modified and scaled the ABE features to the AODV QoS protocol. Then by it seems to be like a cross-layer routing protocol. The MAC layer estimates pro-actively and periodically the available bandwidth of the neighboring links, and the routing layer evaluated the QoS routes complying with the application demands and basing its decision on the MAC layer information.

Route Discovery: -

The aim of the route discovery concept is to discover the route between the sender node and receiver node. And that should meet the constraint applications like bandwidth specified by the application level. So therefore, two flows can transmit to the receiver with different routes and depending upon the network state.

When a source node has to send the data initially it broadcasts RREQ (Route request) to its neighbor. The RREQ packet contains the address of the sender, the channel use, and the requirements at the application level, the destination address and a sequence number. Each mobile node that receives such an RREQ performs an admission control by simply comparing the bandwidth requirement carried in the RREQ packet to the estimated available bandwidth on link it received the RREQ on. It this check goes true(positive), the node adds its own address to the route and forwards the RREQ; otherwise if this check goes true(positive), the node adds its own address to the route and forwards the RREQ; otherwise if the check goes wrong(false) it silently discards the data. This step is different from the other protocols because the admission control is done on the receiver's side not on the emitter's side. This is explained by the fact that in ABE, each node stores the available and evaluated bandwidths of its ingoing links. Finally if the receiver node receives the RREQ, it sends a unicast Route Reply (RREP) to the emitter of the request along the reverse path of the network transmission. The resources are then reserved and stored in the new QoS flow can be then sent in to the network.


By comparing the bandwidth application requirements and the links available bandwidth is not enough to decide about the network ability to convey a flow. In fact, the intra flow contention must be considered when performing the multi hop admission control. The value of Contention Count (CC) can be calculated on a node along a given path. This value is equal to the number of nodes on the multi hop path that are located within the carrier sense range of the considered node. To calculate the CC of each node, the distribution signal power needs to be analyzed. As per the QoS-routing techniques in the multi hop networks and for simplicity reasons the author rather use a direct relation between the end-to-end throughput and the number of hops. After consideration of the intra flow contention on an intermediate node u, which is located at J hops from the source and has received the RREQ from a node i, the available bandwidth considered for admission control, denoted by B (i, u)

B(i, u)=

Where Efinal is the available bandwidth of the link(i, u) as calculated by the ABE protocol.

Article name: QOS Protocols For Estimation Of Resources Computer Science essay, research paper, dissertation