The Congestion Control In Iub Link Of Hsdpa

Published Date: 02 Nov 2017

Abstract: This review paper provides a brief description of the different researches done to tackle the problem of congestion. Congestion as a problem refers to accumulation of packets at the routers, a number of researches have been done to address this problem. The literature survey provided in the following section gives a brief overview about these research papers. In addition to summary, simulation results and analytical output are also presented which on comparison appear very identical to each other.

In the ending section our model is described which utilises and combines the logic elaborated in the literature survey.

Index terms—HSDPA, UMTS, WCDMA, 3GPP,RNC, TCP, UTRAN, QoS, BER, Markov Model, AIMD, etcetra.

Introduction

HSDPA is an upgrade to the 3GPP Rel-99 UMTS technology meeting the ever increasing user requirements. Several enhancements to WCDMA R’99 were considered in 3GPP release 5. Moving the scheduler from the RNC (Radio Network Controller) to the Node B and reducing the Transmission Time Interval (TTI) from 10ms to 2ms allows adapting better to the varying radio channel. The Adaptive Modulation and Coding (AMC) can adapt the modulation rank and coding rate to the current radio channel condition of each user. With a constant transmit power , the effective data rate can so be significantly enhanced for users with good radio channel conditions. The other prominent improvement of the physical layer is the use of Hybrid ARQ (HARQ), where the soft bits from retransmission requested by the UE are combined with soft information from the original transmission prior to the decoding leading to an improved BER performance. In addition to the physical layer improvements, currently the research is focused on enhancements and performance measures on Iub interface. From the operator point of view, the Iub is usually a scarce resource. The HSDPA transport channel, HS- DSCH is controlled by the MAC-hs scheduler which is a new entity located in the Node B. Packet switched traffic via HS-DSCH is scheduled in the Node B and then transmitted over the air interface. Moving the scheduler to the Node B enables more efficient implementation of the scheduling by allowing the scheduler to work with the most recent channel information. The required data for radio channels should arrive with low delay at the Node B side in order to minimize the data buffering in the MAC-hs. The Transport Network Layer (TNL) is optimized by deploying suitable flow control and congestion control techniques minimising the Iub link bandwidth requirements, providing better QoS to the end user and also reducing the operator costs. Compared to Rel-99, HSDPA uses two buffering points at the Node B and at the RNC. Since the transport capacity is limited and the demand of the air interface varies rapidly according to the channel condition, the data flow over the Iub interface should be adapted accordingly. The credit based flow control scheme quantifies the required air interface capacity and informs the RNC as a resource allocation for each user flow separately. However, when the demanded user traffic over the air interface is greater than the available Iub capacity, the congestion occurs at the UTRAN transport. Due to Iub congestion, the packet loss probability increases significantly, causing more retransmissions at higher layers such as RLC and TCP. Hence the offered load is further staggered leading the system into congestion collapse and wasting network resources. To avoid such circumstances, a proper congestion control scheme is required in addition to the credit based flow control mechanism. A credit based flow control mechanism has been developed to optimise the Iub utilisation while providing required QoS to the end user in the HSDPA network. It smoothes down the HSDPA traffic and reduces the burstiness over the Iub interface. The implemented credit allocation algorithm is based on the periodically updated provided bit rate (PBR) of the user-specific priority queues in the Node B. When the Iub link is overloaded, congestion occurs and cannot be protected by the flow control mechanism and a proper TNL based congestion control mechanism is required. 3GPP has introduced two basic mechanisms working at the FP layer (i.e. transport media independent) to perform Iub congestion detection frame loss detection by means of FSN (Frame Sequence Number) supervision and delay build up detection by means of DRT (Delay Relative Time) supervision. In addition to these, a third congestion detection mechanism based on Checksum of HSDPA data frame is considered.

All these algorithms are described briefly in the preceding sections. In addition to this ,a brief description of the analytical and simulation results is also presented.

Overview Of Surveyed Literature

Gap Processing Time Analysis of Stall Avoidance Schemes for HSDPA with Parallel HARQ Mechanisms

The paper[1] was published in 2006. The parallel multichannel stop-and-wait (SAW) hybrid automatic repeat request (HARQ) mechanism is one of key technologies for high-speed downlink packet access in the wideband code division multiple access system. However, this parallel HARQ mechanism may encounter a serious stall problem, resulting from the error of the negative acknowledgement (NACK) changing to the acknowledgement (ACK) in the control channel. Due to this, congestion may take place. This paper investigates 3 stall avoidance techniques the timer-based, the window-based, and the indicator-based schemes, to enhance the MAC layer performance of a parallel multichannel SAW HARQ mechanism adopted in HSDPA. After a thorough analysis and comparison, the indicator based scheme was found to be the best amongst the three.

In this scheme, we make use of two parameters New Data Indicator (NDI) and Transmission sequence number(TSN).NDI is a one bit tag that changes/toggles for each new data and remains the same for old packets and along with TSN we determine the whether Type2 gap has occurred or not . Type2 gap is defined as the lost packet that will never be sent by MAC retransmission scheme due to NACK-TO-ACK error. These situations are examined by extensive simulations.

In addition to simulations, they have also derived probability mass functions and the closed-form expressions for the average gap processing time of these three stall avoidance schemes. The comparison of analytical results and the simulation results are shown below.

Fig. 1. The average gap processing time of the timer-based stall

avoidance scheme with different timer expiration for the 4-channel SAW

HARQ mechanism in the Rayleigh fading channel with Doppler

frequency of 100 Hz.

As seen above the simulation and analytical results are very analogous to each other.

Preventive and Reactive based TNL Congestion Control Impact on the HSDPA Performance

The paper[2] was published in 2008. The main focus of the work presented was to analyze the effect of congestion at the Iub interface on the HSDPA performance. The paper discusses congestion control scheme based on the three congestion detection methods based on the frame sequence number (FSN), Delay Reference Time (DRT) fields of HSDPA data frame and checksum of HSDPA data frame. The simulation results presented in this paper confirm that the congestion in the transport network can be avoided and performance of the network can be significantly improved.

When the Iub link is overloaded, congestion occurs and cannot be protected by the flow control mechanism and hence a proper TNL based congestion control mechanism is required. Both congestion detection and congestion control algorithms are implemented in a cascade with the credits based flow control mechanism which is used to estimate the current air interface capacity in the Node B for HSDPA. The FC and the congestion detection algorithms work independently and provide information to the congestion control module promptly.

Fig. 2. HSDPA MAC-d flow state machine.

At the congestion control state, the MAC-d flow is controlled by the Additive Increase Multiplicative Decrease (AIMD) algorithm. In additive increase, the congestion window is increased by a fixed amount every round trip time. When congestion is detected, the transmitter decreases the transmission rate by a multiplicative factor, for example, cut the congestion window in half when congestion is detected.

Fig. 3. Flowchart of AIMD algorithm.

Mathematical Formula:

w (t+1) = w(t) +1 ; if congestion is not detected.

w (t+1) = w(t) x 0.5 ; if congestion is detected.

In the simulation analysis, the simulation scenarios are configured with the same set of network and air interface parameters in order to validate the effect of the congestion control feature clearly. The simulation results show that for both traffic models (ETSI and FTP traffic) the usage of congestion control algorithms allow to effectively control TNL congestion and minimize the network resource waste due to higher layers retransmissions.

A Unified Approach to Congestion Control

and Node-Based Multipath Routing

The paper[3] was published in 2009. It proposed a framework in which multipath routing and congestion control work in unison to pursue a common objective: the maximization of aggregate utility or surplus over the network. It is a scalable, node-centric alternative to let routers take charge of the multipath function by controlling the split of traffic to each destination among their outgoing links. They proposed a combination of gradient-based multipath routing with primal and dual congestion control. They also demonstrated through ns2-simulations, the collective behavior of the system, in particular, that it reaches the desired equilibrium points.

A Markovian Model for HSDPA TNL Congestion

Control Performance Analysis

This paper[4] was published in 2008,it aims at solving the congestion problem by using CC algorithms and probability models. The congestion control is done per MAC-d flow in the network. There are three detection algorithms that work independently and detection signals are forwarded to the congestion control module in the RNC. The CC module takes appropriate actions to reduce the data rate of the users who are causing congestion for the TNL network. As soon as the congestion is detected, the CC module immediately reduces the data rate by a factor β (configurable parameter), and then increases the data rate stepwise. This phase of data rate increasing is known as the recovery phase. Here the data rate is increased using the AIMD algorithm which uses a principle similar to that used in TCP. There can be several consecutive congestion indications one after the other causing a data rate reduction by the factor β at each time. Under heavy congestion situations such consecutive events can lead even to a low data rate for each flow . Its most prominent feature is the use of probability models to predict the impending congestion. The probability model/transition model is formed by analysing the no. of congestion indications or CI ,time interval between two CI which is denoted by states and the step size. The arrival process of the packets can be modelled as a discrete-time process. Each CI arrival causes the current data rate to reduce by a factor β. The state is defined as the transmission rate immediately after the CI arrival which means just after the CC decision has been taken. The discrete state random process is denoted by X(t).

X ( t ) = X1, X2, X3 ,……, Xn for all n ---------(1)

The discrete state space or the set of possible values that X(t) can take ranges from 0 to m, where m corresponds to the maximum possible data rate that occurs immediately after a single CI event. Between two CI arrivals, the data rate increase at a constant rate per unit time or step. The steps and states as used in the analytical model are shown in Fig. 4, where the Y axis shows the transmission rate and the X-axis shows the elapsed time. The inter-arrival time between the arrivals of the nth and the (n+1)th CI signals is denoted by a random variable Tn. So that the inter-arrival distribution, can be defined as follows:

An(t) = P[Tn≤ t] --------------------- (2)

Fig. 4. State and step representation.

Within a step time or a unit time, there is a finite probability for more than one CI triggers. Such multiple CI events occurring within the same time step are modelled as simultaneous CI arrivals at the end of the corresponding step where the CC action is taken. Those multiple CI arrivals have multiple drops at the same time and are modelled as multiple departures. In order to model them analytically the multiple transition probabilities are taken into account not in the inter-arrival time distribution of the CIs but separately in the calculation of transition probabilities.

The complete behaviour is described by the Discrete-time Semi-Markov Process with Multiple Departures or Embedded Markov Chain with Multiple Departures (EMC-MD).The analytical throughput and the simulated throughput are graphically illustrated below.

Fig. 5. Simulation results: transmission rate.

Both simulation and analytical results are agreeing with an error of 12%, calculated with reference to the simulation results.

Table II: Comparison of throughput values

Simulation

Analytical

Average

Throughput

590 kbps

660kbps

TCP Congestion Control over HSDPA:

an Experimental Evaluation

The paper[5] was published in 2012. In this paper, they have considered four TCP congestion control variants: TCP NewReno, which is the only TCP congestion control standardized by IETF, TCP BIC and TCP Cubic, which have been selected as default congestion control algorithms in the Linux OS, and TCP Westwood+. Regarding TCP performances, authors reported measurements of good put, retransmission percentage and excess one-way delay. The User Equipment (UE), which is a mobile phone equipped with a commercial HSDPA card, was tested in a static scenario so that handovers don’t occur during measurements. Many experimental runs were carried out, considering different TCP flows and a rich set of metrics such as good put, throughput, round trip time, number of time outs and packet loss ratio were evaluated for different TCP variants.

Fig. 6. CDFs of the packet retransmission percentage in the case of one flow sharing the HSDPA downlink.

Fig. 7. CDFs of the RTT in the case of one flow sharing the HSDPA

downlink.

Fig. 8. CDFs of the aggregated goodput in the case of one flow sharing the

HSDPA downlink.

TCP Westwood+ was found to be particularly effective

over wireless links.

CONCLUSIONS

After a thorough study of the material presented above, we conclude that, in [1] there can be congestion due to a different case altogether, conversion of NACK to ACK due to some problem while transmission of the signal , which is highlighted by this paper. In addition to that they have suggested three solutions to this problem along with their pros and cons.

In [2], a very powerful probability algorithm is provided, which utilises markov chains to predict the impending congestions. The results described in this paper show a marginal difference between the simulated and analytical model. It gave us the idea of adjusting the data rate on runtime by a factor β.

The algorithms presented in [3], analyse both link level and end user level performance. Preventive part of their algorithm prevents the congestion to occur altogether, and when preventive part fails, the reactive part of the algorithm comes into action to handle the congestion.

We will develop an algorithm which takes into account the advantages of all these algorithms, that is, from [1], our algorithm will continuously be active to monitor any change in acknowledgments, using [3] the preventive and reactive scheme will start on the onset of congestion, and finally from [2] the factor β will be changed according to the buffer space available at runtime.