Real-Time Communication in Unconstrained Shared Ethernet Networks:
The Virtual Token-Passing Approach
Francisco Borges Carreiro
1, 2, Ricardo Moraes
3, José Alberto Fonseca
2and Francisco Vasques
3 1Department of Eletro-Eletrônica - CFET-MA, 65025-001 - São Luís Maranhão - Brasil
2Department of Electronics and Telecommunications, Univ. of Aveiro, 3810-143 Aveiro, Portugal
3
Department of Mechanical Engineering, University of Porto, 4200-465 Porto, Portugal
fborges@ieeta.pt, rmoraes@fe.up.pt, jaf@det.ua.pt, vasques@fe.up.pt
Abstract
Traditional shared Ethernet networks have a non-deterministic behaviour, which impairs the support of real-time communication. In this paper, it is proposed a shared Ethernet deterministic architecture, able to interconnect sensors, controllers and actuators at the field level, allowing the coexistence of standard devices with enhanced (real-time) devices. Such solution is based on the control of the medium access right, by means of a virtual token passing procedure among enhanced stations, complemented by an underlying prioritization mechanism. Such underlying mechanism guarantees that, whenever an enhanced (real-time) station is contending for the bus access, it will be able to access the bus prior to any other station. Thus, it enables the traffic separation between standard and enhanced (real-time) stations, being able to guarantee real-time communication in unconstrained traffic environments.
1. Introduction
Simplicity, high speed and low cost are the main advantages that are imposing Ethernet as the preferred communication protocol to support industrial control applications [1]. Such simplicity derives from its Medium Access Control protocol, which is based on the collision detection between randomly initiated transmissions. Whenever a collision is detected, a distributed probabilistic algorithm is initiated to solve the serialization problem between the contending messages. Such algorithm, which is based on the local knowledge of occurred collisions, implements a decentralized Medium Access Control protocol. Nevertheless, one of its main disadvantages is the inherent non-determinism of the probabilistic contention resolution. Therefore, the shared Ethernet standard is not able to provide a real-time service to the supported applications, unless additional functionalities are introduced.
Several approaches and techniques have been developed to provide real-time behaviour to Ethernet-supported applications. However, few of those techniques allow standard devices to coexist with enhanced stations in the same network segment. Relevant exceptions [2][3] have strong limitations related to the number of allowed real-time stations [2] or the requirement for the use of specific hardware [3].
The main target of this paper is to propose a shared Ethernet deterministic architecture, able to interconnect sensors, controllers and actuators at the field level, allowing the coexistence of Ethernet standard devices with multiple enhanced (real-time) devices in the same network segment.
Such solution is based on the control of the medium access right, by means of a virtual token passing procedure among the enhanced (real-time) stations, complemented by an underlying prioritization mechanism which guarantees that, whenever an enhanced station is contending for the bus access, it will be able to access the bus prior to any standard station. Such underlying mechanism enables the traffic separation between standard and enhanced (real-time) stations, and therefore is able to guarantee real-time communication in unconstrained traffic environments.
This paper is organized as follows: section 2 describes the h-BEB (high priority BEB) collision resolution algorithm [2], as well as the VTPE token passing procedure. Section 3 shows how the combination of VTPE and h-BEB algorithms leads to a protocol where enhanced stations are able to transfer real-time traffic within unconstrained traffic environments. A temporal analysis of this new protocol is presented in section 4, showing that the timings are adequate for the target applications. In section 5, it is presented the state-of-the-art for the real-time communication support with shared Ethernet networks. Finally the paper is concluded in section 6.
2. Background
The proposed architecture is based upon both the h-BEB collision resolution algorithm [2] and the VTPE token passing algorithm . Therefore, in this section we briefly describe these two algorithms.
2.1. The high priority Binary Exponential Backoff Algorithm (h-BEB)
The CSMA/CD (Carrier Sense Multiple Access with Collision Detection) protocol is the protocol implemented at the MAC layer of both ANSI/IEEE 802.3 [4] and Ethernet local area networks. For a 10/100 Mbps Ethernet implementation, the following set of parameters is used:
Table 1: Ethernet parameters.
Parameters Values
SlotTime 512 bit times 64 byte times InterFrameGap 96 bit times 12 byte times JamSize 32 bit times 4 byte times MaxFrameSize 12144 bits 1518 bytes MinFrameSize 512 bits 64 bytes
Basically, the CSMA/CD protocol works as follows (Figure 1a): when a station wants to transmit, it listens to the transmission medium. If the transmission medium is busy, the station waits until it goes idle; otherwise, it transmits immediately. If two or more stations simultaneously begin to transmit, the transmitted frames will collide. Upon the collision detection, all the transmitting stations will terminate their own transmission and send a jamming sequence to ensure that all the transmitting stations abort the
transmission1. When the transmission is aborted due to
a collision, it will be repeatedly retried after a randomly evaluated delay (backoff time) until it is, either successfully transmitted, or definitely aborted (after a maximum number of 16 attempts) [4].
The backoff delay is evaluated by locally executing the Binary Exponential Backoff (BEB) algorithm. Such algorithm operates as follows: after the end of the jamming sequence, the time is divided into discrete
slots, whose length is equal to the slot time2. The
backoff time is given by tbackoff = r×T, where r is a
random integer in the range , k is the
smaller of n or 10 (n is the number of retransmission attempts) and T is the slot time in seconds. This means
that the station will wait between 0 and 2n–1 slot times,
being n the number of collision resolution rounds. After 10 attempts, the waiting interval is fixed at 1023
1 2 0≤r≤ k −
slot times, and finally after 16 attempts the transmission is discarded. Transmit Frame Bus Busy? Colision Detect Start Transmission Wait Finish yes yes no Transmission Done? no send jam Done: TransmitOk yes Increment Attempts to many attempts? wait backoff time compute backoff Done: ExcessiveCollision Error yes no no Transmit Frame Bus Busy? Colision Detect Start Transmission Wait Finish yes yes no Transmission Done? no send jam Done: TransmitOk yes Increment Attempts to many attempts? Done: ExcessiveCollision Error yes no no (a) (b)
1 More accurately, when detecting a collision, the station always
finishes the transmission of the Preamble and the Start of Frame Delimiter (64 bits). Afterwards, it transmits a jamming sequence (32 bits), and then stops.
2 For Ethernet and Fast Ethernet (10/100 Mbps) networks, one slot
time is the time required for transmitting the minimum frame size (512 bits), that is, respectively, 51.2 and 5.12 µsec.
Figure 1. CSMA-CD protocol with BEB resp. h-BEB collision resolution algorithms.
On the other hand, a station implementing the h-BEB algorithm operates as follows (Figure 1b): whenever there is a collision, the station starts immediately transmitting (backoff interval equal to zero). That is, the h-BEB modification sets the backoff delay parameter to 0 in the h-BEB station. This behaviour guarantees the highest transmitting probability to the h-BEB station, as it will always try to transmit its frame in the first slot, while all the other
stations will wait between 0 and 2n-1 slot times.
The h-BEB collision resolution algorithm can be used to support real-time traffic separation, as the traffic generated by the h-BEB station will be always transferred prior to the traffic generated by the other stations.
2.2. The Virtual Token-Passing Ethernet
A VTPE system consists of a shared Ethernet bus, where all the station nodes are interconnected by a logical ring . All the producer stations have an internal mechanism that enables them to verify if the moment to access the bus has arrived. This internal mechanism is triggered by the reception of a VTPE frame transmitted by other node or by the elapsing of time, if the bus keeps being idle. There is thus no explicit token passing between the producer stations. VTPE allows the inclusion of an indeterminate number of stations that just receive the frames. Figure 2 shows a block diagram of the VTPE algorithm, which will be explained below.
Each producer station in a VTPE system has a node address (NA), which can be defined between 1 and np,
np being the number of active producers in the system.
which identifies the node that can access the bus in a specific time interval. When the node address (NA) in a node is equal to the access counter (AC) value, this node is authorized to transmit a frame.
Figure 2. VTPE flow chart.
The AC is incremented a short time (t1) after an
interrupt signalling the reception of a VTPE frame has occurred. The AC is also incremented if the bus is felt
idle during a pre-defined interval called t2. The first
situation occurs when the previous node holding the implicit token has transmitted a frame. The second situation occurs when the previous node holding the token has nothing to transmit and thus the bus stays
idle. The t1 time is just used to enable the slowest
processor in the system to decode the VTPE frame (read the frame). When the AC equals the number of producers np, its value is reset to 1.
If the producer nodes have nothing to transmit during a long time interval then the bus stays idle and the stations can become unsynchronised due to the absence of the synchronization event represented by the frame reception interrupt. The VTPE protocol has a mechanism to overcome this problem, which is based in the IBC (idle bus counter).
If the bus is idle all producers increase the IBC and compare it with a predefined value, K. The maximum
interval without bus activity is then bounded to K×t2. If
the IBC becomes greater than K, the node that holds the token must send immediately a special frame to synchronize the access counters. The use of the condition IBC ≥ K instead of only IBC = K solves the problem of an eventual absence of the node that would be holding the token when IBC = K. The IBC is reset whenever a frame has been received.
3. The VTPE/h-BEB Architecture
The main target of the VTPE/h-BEB architecture is to enable the coexistence of unconstrained Ethernet stations within a real-time communication infrastructure, by means of traffic separation mechanisms. Such traffic separation mechanisms guarantee that, whenever a VTPE/h-BEB station holding the token has real-time traffic to transfer, it will win the collision resolution procedure and will transfer its message before any other message from non real-time stations.
Interrupt IBC=0 Set timer w ith t1
t1 expires Increase AC AC=NA? no yes Authorize a frame transmission Set timer w ith t2
t2 expires Is the bus idle? yes no Wait for interrupt Increase IBC IBC>=k? yes no
The node, w hich holds the token sends a synchronization frame
Waits for interrupt LEGEND
IBC = Idle Bus Counter AC = Access Counter NA = Node Address
Basically, the VTPE/h-BEB architecture implements a Virtual Token Passing procedure [5], on top of the h-BEB collision resolution algorithm [2]. The underlying h-BEB algorithm enables the VTPE procedure, running between the VTPE/h-BEB stations, to coexist with Ethernet devices with unconstrained traffic. By circulating a virtual token among all the h-BEB (real-time) stations, it restricts the transmission of h-BEB messages from just one station, at each instant. The previous limitation of the h-BEB algorithm, i.e., just one h-BEB station per network segment, is thus circumvented. Figures 3 and 4 illustrate the proposed VTPE/h-BEB architecture.
HUB IBM Compatible IBM Compatible IBM Compatible IBM Compatible IBM Compatible IBM Compatible Virtual Ring 1 - VTEPE/hBEB station 2 - Ethernet station 1 1 1 2 2 2
Figure 3. VTPE/h-BEB system architecture.
A station implementing the VTPE/h-BEB algorithm operates as illustrated in Figure 4. Whenever a frame finishes to be transferred, an interrupt occurs simultaneously in all nodes. Therefore, the interrupt event is used to synchronize the AC counters. Whatever the VTPE/h-BEB station, when its Access Counter (AC) is equal to the Node address (NA), it means that the station is holding the token. If the station has something to transmit, the h-BEB algorithm will immediately start, guaranteeing that the station will win the medium access.
If the station holding the virtual token does not have any message to be transferred, it will allow Ethernet standard stations to contend for accessing the bus,
during a time interval t2. If the bus remains idle during
all the AC counters will be incremented, which corresponds to an implicit token passing.
Figure 4. Control Flow Summary – VTPE/h-BEB.
If an Ethernet standard station tries to transmit
during the time interval t2, two different situations can
arise: either the message is normally transmitted or a collision resolution procedure starts. If a transmission occurs, then the algorithm just waits for the interrupt at the end of the message transfer. If a collision resolution starts, then it can be either generated just by standard Ethernet stations or it can also include an active h-BEB station holding the token. The first scenario, i.e., a collision involving just standard Ethernet stations, can be detected if a bus idle occurs with duration greater than 51,2 µsec (slot time duration at 10Mbps). In such case, the VTPE/h-BEB stations can pass the virtual token as the h-BEB station that is holding the token has nothing to transmit.
As deduced in [2], the h-BEB algorithm solves collisions in a bounded time, or it eventually discards the message. This enables the definition of a time
interval t3, greater than the h-BEB collision resolution
interval. If a message does not start to be transferred
during the t3 interval, then a collision between
messages from multiple standard Ethernet stations has occurred (as the h-BEB collision resolution algorithm
would have succeeded during that interval). If the t3
interval expires, it is then possible to pass again the token and thus an interrupt is generated.
4. Timing Analysis
In this section, it is presented the timing analysis of an Ethernet network interconnecting multiple VTPE/h-BEB stations with Ethernet standard stations. This analysis clearly illustrates the real-time behaviour of the proposed VTPE/h-BEB architecture.
interrupt AC=NA has message to be transmitted? increment AC yes start h-BEB algorithm h-BEB finished? no set t2 bus idle t2 expired? resett2 yes no yes no t3 expired? message being transferred? message tranf. finished? yes no set t3 yes no no bus idle during more than 9.6 ms yes no yes yes no yes no
Consider a network with n VTPE/h-BEB stations, with addresses ranging from 1 to n. Each VTPE/h-BEB station accesses the network according to the VTPE/h-BEB scheme, i.e., first station 1, then station 2, 3,… until station n, and then again station 1, 2, …n. The standard Ethernet stations implement the traditional BEB collision resolution algorithm.
First of all, consider a two-collision scenario. In such case, the maximum delay to transfer a real-time message, when the VTPE/h-BEB station is holding the token, is illustrated in Figure 5. According to the VTPE/h-BEB scheme, such station transmits its message using the h-BEB algorithm; that is, it always tries to transmit its message in the first time slot.
t0 80 collision collision I1 PA J1I2 PA J2 80 t
Figure 5: A 2-collision scenario solved by the h-BEB collision resolution algorithm.
Therefore, when a VTPE/h-BEB station holding the
token has a message ready to be transferred (PA), it will
wait an Inter Frame Gap (I1: 12 byte times) before
starting to transmit. If a collision occurs during the
transfer of the first 64 bytes of message PA, a jamming
sequence will be broadcasted (J1: 4 byte times).
Afterwards, the station will wait again during an Inter
Frame Gap (I2: 12 byte times) and, according to the
h-BEB algorithm, it will immediately start to transmit its message. If a second collision occurs, a new
jamming sequence (J2) will be broadcasted and station
A will wait again for the Inter Frame Gap (I3), before
starting to transmit. The cumulative result (from t0 up
to the beginning of the third attempt) is 160 bytes or 0,128 ms (at a 10 Mbps bit rate). The maximum time that a VTPE/h-BEB station holding the token will wait before starting to transfer a message or eventually to discard it is 0.960 ms (Table 2).
It is clear that the h-BEB algorithm solves collisions in a bounded time, or it eventually discards the message. Therefore, it is of utmost importance to focus on the probability of a message frame being discarded by the h-BEB algorithm, whenever the number of collision resolution rounds exceeds 15.
Table 2: Maximum delay to start transferring a message: h-BEB collision resolution algorithm.
Retry
Number Max delay (# slots) Max cumulative delay (# slots) Max delay (ms)
1 1 1 0,064 2 1 2 0,128 3 1 3 0,192 … 9 1 9 0,576 10 1 10 0,640 … 14 1 14 0,896 15 1 15 0,960 16 discard frame
Such probability has been analytically evaluated for an highly loaded network scenario, and is equal to
1,22×10-4 and 1,95×10-3, respectively for small (5
stations) and large population (65 stations) scenarios. For more realistic load scenarios, it has been verified by simulation that a h-BEB station never discards any packet, whatever the simulated network load (simulation scenario: 75×104 h-BEB simulated messages in a 10Mbps network with 64 standard Ethernet stations and one h-BEB station, with a network load ranging from 40% to 110%) [6]. Such results are consistent with the claim that the h-BEB algorithm is able to support most part of the soft real-time applications, as they confirm a rather small probability of any message being discarded.
Therefore, if it is considered that no message is discarded by the VTPE/h-BEB station holding the token, the maximum time that a VTPE/h-BEB station holding the token waits to transfer a real-time message is given by:
message col
hBEB t IFG t
T = + + (1)
where tcol is the worst-case delay to start transferring a
message (0.960 ms), IFG is the Inter Frame Gap (12
byte-times) and tmessage is the time to transfer a message
from the VTPE/h-BEB station, which is the maximum message length.
On the other hand, when the VTPE/h-BEB station holding the token does not have any real-time message ready to be transferred, the standard Ethernet stations in the network segment can try to start transferring their own messages. In such case, all the VTPE/h-BEB
stations will wait during a time interval t2, within
which any Ethernet standard station may try to start transferring a message. If the collision resolution round
is longer than t3 (0,96ms), or if the bus remains idle
during a time interval equal to t2, an interrupt will be
generated and all the AC counters will be incremented (i.e., there will be a Virtual Token Passing).
In Figure 6 it is exemplified the maximum time interval that a VTPE/h-BEB station is allowed to hold
the token, even if it does not have any real-time message ready to be transferred.
Such worst-case arises when multiple collisions
occur. In such case the time interval t3 must be long
enough to allow an h-BEB message transfer, as the VTPE/h-BEB stations that are not holding the token do not known if the colliding messages are from just Ethernet standard stations, or if there is also a message from a VTPE/h-BEB station holding the token. In the
latter, the time interval t3 guarantees that the
VTPE/h-BEB station holding the token station will be able to transmit its message and an interrupt will occur when the message transfer is finished. Otherwise, if the collision resolution is not solved during the time
interval t3, it means that the collisions are occurring
just among standard Ethernet stations. Therefore, an
interrupt will be generated after t3 and the next
VTPE/h-BEB station in the logical ring will be able to contend for the medium access.
The worst-case for the token holding time occurs
when at instant (t3 - ε), a standard Ethernet station
starts to transmit a 1518-byte message (tover), which is
the longest message that can be transferred in an Ethernet network. t3 expires tover ε t3 t t0 1518 bytes
Figure 6: Token holding time.
Therefore, the maximum time that a VTPE/h-BEB station may hold the token is given by:
over
TH t t
T = 3+ (2)
As the token rotation time is the time interval between two consecutive token visits to a particular station, the worst-case token rotation time, denoted as
TRT, is given by:
TH
T n
TRT= × (3)
where TTH is as defined in Equation (2). The value TRT
represents the worst-case time interval between two consecutive token arrivals to any VTPE/h-BEB station (m=1,…, n).
5. Real-time behaviour in Ethernet
Several approaches and techniques have also been developed to provide a real-time behaviour to Ethernet-supported applications. Some of these approaches are based on the Switched Ethernet technology, where the hub-based technology is replaced by switches, trying to eliminate CSMA/CD collisions by dividing “collision
domains into point-to-point connections” [7]. The Switched Ethernet standard, introduced in the early 90s (IEEE 802.1D), enables the micro-segmentation of the network by regenerating information only to the receiving port. Additionally, when using Ethernet Switching Hubs, it is possible to manage network traffic, by means of the adequate setting of data flow permissions and priorities. Also, adequate network management functionalities were specified both by the IEEE 802.1p and IEEE 802.1q VLAN standards, in order to enhance the transmission of critical data. The latter extends the priority handling aspects of the 802.1p standard, by providing space in the VLAN Tag to indicate traffic priorities (up to 8 distinct traffic classes) to support virtual local area networks (VLANs). Nevertheless, avoiding collisions does not guarantee a real-time behaviour in an unconstrained environment, as the congestion/overflow of the switches may still occur.
When considering, as in this paper, the traditional shared Ethernet standard, different approaches can be identified in what concerns the support of real-time communications [2, 8]: ensuring a deterministic
collision resolution, by modifying the collision
resolution algorithm or avoiding collisions, by controlling the medium access rights of each station. A third approach (that is not deterministic) is to reduce the number of occurring collisions, enhancing the network responsiveness to real-time message requests.
To ensure a deterministic collision resolution it is required the modification of the MAC sub-layer of all the Ethernet stations in the network segment, to guarantee that the colliding frames are serialized in an upper-bounded time interval. One of the first proposals was made by Takagi et al. [9], which presented a CSMA/CD protocol with deterministic contention resolution (DCR). In the absence of collisions, the CSMA/DCR protocol implements the CSMA/CD access method. In collision situations, a binary search tree is used to sort the colliding nodes. A priority hierarchy is enforced, i.e., higher priority nodes try to access the medium prior to the lower priority ones, using an implicit token passing mechanism. The DOD-CSMA-CD protocol [10] improved the CSMA-DCR protocol. It uses network station indices computed on-line, rather than pre-assigned. The drawbacks are the need for MAC modification and a high worst-case transmission time, when compared with the average.
For controlling the medium access right many approaches have been proposed. One of the first was the one by Chen and Lu [11] based on the TDMA (Time Division Multiple Access) paradigm, where each station has a pre-allocated transmission time interval. Another approach was proposed by Pritty et.
al. [12], based on the use of the Timed Packet Release
principle, where a monitor node periodically transmits a slot pulse to synchronize the medium access.
Some recent solutions use master-slave techniques, where a special node, the Master, instructs the other nodes, the slaves, to transmit in specific instants. Collisions are thus avoided. Pedreiras et. al. [13] proposed a master-multi slave technique called FTT-Ethernet (Flexible Time-Triggered FTT-Ethernet) to schedule communications in a shared Ethernet network.
Another way to provide a deterministic
collision-free environment is to use an explicit token passing
procedure, where each station is allowed to access the medium only during the token holding intervals. Venkatramani and Chiueh [14] proposed the RETHER (real-time Ethernet) protocol, where the network operates in the CSMA mode until a real-time request arrives, passing then to the RETHER mode, where all nodes operate according to a token passing protocol. In this mode time is divided into cycles during which the token regulates the access to the bus. The token visits first all real-time nodes and, if time is left during the cycle, it visits the non-real time nodes.
In [15], J. Lee et al. proposed the use of the IEEE 802.4 Token-Passing Bus Access method directly on top of the Ethernet Physical Layer, where a specifically proposed service translator performs the required translation of frame formats and interface functions.
Other token-passing proposal based in the explicit token-passing is named RT-EP (Real-Time Ethernet Protocol). In RT-EP the access to the bus is carried in two phases, arbitration and application message transmission. In the first, the token visits all nodes to determine the highest priority message and is after sent to the correspondent node for transmission (application message transmission phase). After, the same node starts a new arbitration phase.
In what concerns solutions allowing standard devices to coexist with enhanced modified stations in the same network segment, imposing higher priority to privileged traffic, there are not many works to report. Besides the h-BEB algorithm proposed by Moraes and Vasques [2] and described in section 2, one can identify the EquB protocol proposed by Sobrinho and Krishnakumar. It enables a privileged access to real-time traffic with a FCFS (First-Come-First-Served) discipline. The collision resolution mechanism for real-time sources requires the disabling of the exponential back-off mechanism and the transmission of jamming sequences with durations dependent on the contention periods experienced by the real-time traffic.
The third approach to support real-time communica-tions in shared Ethernet environments is to reduce the number of occurring collisions, enhancing the network responsiveness to real-time message requests. Note that these solutions are non-deterministic, as collisions are still solved in a probabilistic way.
Molle and Kleinrock [16] proposed a CSMA algorithm, called Virtual Time CSMA (VTCSMA). It
uses a probabilistic approach combined with specific timing parameters (arrival time, laxity, deadline, length) for the collision resolution, enabling different scheduling policies. Zhao and Ramamritham [17] presented a performance analysis of the four VTCSMA protocols: VTCSMA-A, -T, -D and -L, which implement the arrival-time-first, minimum-transmission-time-first, minimum-deadline-first, and minimum-laxity-first policies, respectively.
Another relevant modification proposed to the CSMA/CD protocol is the Window Protocol [18-20], which implements a dynamic time window to reduce the number of occurring collisions. It operates as follows: when just one host has a message ready to be transmitted, if the message is within the window, then it will be sent; if several hosts have messages to be transmitted within the window, the window size is reduced according to the selected policy, until there is just one remaining message within the window; if there are no nodes with messages within the window, the window size can be increased.
In [21] is presented the Dynamic pi persistent
CSMA/CD protocol. It is similar to the p-persistent protocol, but with a transmission probability that depends on the laxity of the ready packet. It also implements a time window to reduce the number of collisions in heavily loaded systems.
Molle et al. [22] proposed a BEB compatible algorithm, the Binary Logarithmic Arbitration Method (BLAM), with a modified collision counter policy. According to Christensen [23], following a successful transmission, all the stations will have an equal access probability to the medium. Therefore, it eliminates the packet starvation effect [24]. The Capture Avoidance Binary Exponential Backoff (CABEB) algorithm proposed by Ramakrishnan and Yang [25] addresses also the packet starvation effect. It enhances the collision resolution algorithm for the special case when a station attempts to capture the channel following an uninterrupted sequence of message transfers.
Finally, another approach is to use traffic smoothing mechanisms, introduced by Kweon et al. [26][27], where the packet generation rate (from the upper layers) is kept below a defined threshold, the network-wide input limit. Several policies for traffic smoothing have been proposed. The HIMD (Harmonic-Increase and Multiplicative Decrease) [26] uses the credit bucket depth and the refresh period as dynamic traffic regulator; in the absence of collisions, it periodically increases the input bound through periodically reducing the refresh period. In [28] the smoothing actions are performed by a fuzzy controller, where the network load is observed along determined time intervals, via the throughput measurement and the number of occurring collisions. Finally, in [29] a middleware system is proposed to regulate the network access by means of a polling mechanism.
6. Conclusions
As referred before, the major motivation of this paper was to “propose a solution enabling the support of real-time communications in shared Ethernet environments, where unconstrained Ethernet standard devices can coexist with multiple enhanced devices”.
To address this problem, it was proposed a solution based on the Virtual Token-Passing procedure, where an underlying high priority Binary Exponential Backoff (h-BEB) algorithm [2] guarantees the medium access right to the VTPE station that is holding the token. This allows Ethernet standard devices to coexist with multiple VTPE enhanced stations, imposing a higher priority for the transfer of VTPE/h-BEB related traffic and guaranteeing the required traffic separation.
The timing analysis included in the paper shows that, for a moderate number of nodes, token rotation time of the order of several milliseconds can be obtained. This figure seems adequate for real-time applications in the automation domain.
References
[1] J.-D. Decotignie, "A perspective on Ethernet-TCP/IP as a fieldbus," presented at Proceedings of LORIA. 4th International Conference on Fieldbus Systems and their Applications, 15-16 Nov. 2001, Nancy, France, 2002. [2] R. Moraes and F. Vasques, "A Probabilistic Analysis of
Traffic Separation in Shared Ethernet Systems Using the h-BEB Collision Resolution Algorithm," presented at 13th International Conference on Real-Time Systems - RTS'2005, Paris - France, 2005.
[3] J. L. Sobrinho and A. S. Krishnakumar, "EQuB - Ethernet quality of service using black bursts," presented at 23rd Conference on Local Computer Networks, Boston, MA, USA, 1998.
[4] "IEEE standards for local area networks: carrier sense multiple access with collision detection (CSMA/CD) access method and physical layer specifications," in
ANSI/IEEE Std 802.3-1985, 1985.
[5] F. Carreiro, J. Fonseca, V. Silva, and F. Vasques, "The Virtual Token Passin Ethernet: Implementation and Experimental Results," presented at Proceedings of the 3rd Workshop on Real-Time Networks, Catania, Italy, 2004.
[6] R. Moraes and F. Vasques, "Real-Time Traffic Separation in Shared Ethernet Networks: Simulation Analysis of the h-BEB Collision Resolution Algorithm," presented at 11th IEEE International Conference on Embedded and Real-Time Computing Systems and Applications, Hong Kong - China, 2005.
[7] E. Jasperneite and P. Neumann, "Switched Ethernet for factory communication," presented at ETFA 2001. 2001 8th International Conference on Emerging Technologies and Factory Automation. Proceedings, 15-18 Oct. 2001, Antibes-Juan les Pins, France, 2001. [8] P. Pedreiras, L. Almeida, and J. A. Fonseca, "The Quest
for Real-Time Behavior in Ethernet," in Handbook of
Industrial Information Technology, R. Zurawski, Ed.:
CRC Press, 2005, pp. 48-1, 48-14.
[9] A. Takagi, S. Yamada, and S. Sugawara, "CSMA/CD with Deterministic Contention Resolution," Selected
Areas in Communications, IEEE Journal on, vol. 1, pp.
877-884, 1983.
[10] G. Le Lann and N. Rivierre, "Real-Time Communications over Broadcast Networks: the CSMA-DCR and the DOD-CSMA-CD Protocols," INRIA 1863, 1993.
[11] W.-H. Chen and C.-C. Lu, "CSMA/CD/TDMA: A dynamic combination for voice and data integration," presented at Proceedings of IEEE INFOCOM '90: Ninth Annual Joint Conference of the IEEE Computer and Communications Societies, Jun 3-4 1990, San Francisco, CA, USA, 1990.
[12] D. W. Pritty, J. R. Malone, D. N. Smeed, S. K. Banerjee, and N. L. Lawrie, "A real-time upgrade for Ethernet based factory networking," presented at Proceedings of IECON '95 - 21st Annual Conference on IEEE Industrial Electronics, 6-10 Nov. 1995, Orlando, FL, USA, 1995.
[13] P. Pedreiras, L. Almeida, and P. Gai, "The FTT-Ethernet protocol: Merging Flexibility, Timeliness and Efficiency," presented at Proceedings of the 14th Euromicro Conference on Real-Time Systems, Austria, 2001.
[14] C. Venkatramani and T.-c. Chiueh, "Supporting real-time traffic on Ethernet," presented at Proceedings Real-Time Systems Symposium, 7-9 Dec. 1994, San Juan, Puerto Rico, 1994.
[15] J.-Y. Lee, H.-J. Moon, S. Moon, W. Kwon, S. Lee, and I. Park, "Token Passing Bus Access Method on the IEEE 802.3 Physicl Layer for Distributed Control Networks," presented at Proceedings of the 15th IFAC Workshop on Distributed Computer Control Systems, Italy, 1998.
[16] M. Molle and L. Kleinrock, "Virtual Time CSMA: Why Two Clocks Are Better than One," IEEE Transactions
on Communications, vol. COM-33, pp. 919-933, 1985.
[17] W. Zhao and K. Ramamritham, "Virtual time CSMA protocols for hard real-time communication," IEEE
Transactions on Software Engineering, vol. SE-13, pp.
938-952, 1987.
[18] J. F. Kurose, M. Schwartz, and Y. Yemini, "Multiple-access protocols and time-constrained communication,"
ACM Comput. Surv., vol. 16, pp. 43-70, 1984.
[19] J. F. Kurose, M. Schwartz, and Y. Yemini, "Controlling window protocols for time-constrained communication in multiple access networks," Communications, IEEE
Transactions on, vol. 36, pp. 41-49, 1988.
[20] W. Zhao, J. A. Stankovic, and K. Ramamritham, "A window protocol for transmission of time-constrained messages," Computers, IEEE Transactions on, vol. 39, pp. 1186-1203, 1990.
[21] R.-H. Jan and Y.-J. Yeh, "CSMA/CD protocol for time-constrained communication on bus networks," IEE
Proceedings I (Communications, Speech and Vision),
vol. 140, pp. 197-202, 1993.
[22] M. L. Molle, "A New Binary Logarithmic Arbitration Method for Ethernet," Computers Research Institute, University of Toronto, Toronto Canadá CSRI-298, July 1994.
[23] K. J. Christensen, "Performance evaluation of the binary logarithmic arbitration method (BLAM)," presented at Proceedings of LCN - 21st Annual Conference on Local Computer Networks, 13-16 Oct. 1996, Minneapolis, MN, USA, 1996.
[24] B. Whetten, S. Steinberg, and D. Ferrari, "The packet starvation effect in CSMA/CD LANs and a solution," presented at Proceedings of 19th Conference on Local Computer Networks, 2-5 Oct. 1994, Minneapolis, MN, USA, 1994.
[25] K. K. Ramakrishnan and H. Yang, "Ethernet capture effect: Analysis and solution," presented at Proceedings of the 19th Conference on Local Computer Networks, Oct 2-5 1994, Minneapolis, MN, USA, 1994.
[26] S.-K. Kweon, K. G. Shin, and G. Workman, "Achieving real-time communication over Ethernet with adaptive traffic smoothing," presented at RTAS 2000: 6th IEEE Real-Time Technology and Applications Symposium, May 31-Jun 2 2000, Washington, DC, USA, 2000. [27] S.-K. Kweon, K. G. Shin, and Z. Zheng, "Statistical
real-time communication over Ethernet for manufacturing automation systems," presented at Proceedings of RTAS'99: Fifth IEEE Real-Time Technology and Applications Symposium, 2-4 June 1999, Vancouver, BC, Canada, 1999.
[28] A. Carpenzano, R. Caponetto, L. Lo Bello, and O. Mirabella, "Fuzzy traffic smoothing: an approach for real-time communication over Ethernet networks," presented at Proceedings 4th IEEE International Workshop on Factory Communication Systems, 28-30 Aug. 2002, Vasteras, Sweden, 2002.
[29] V. Liberatore, "Scheduling of network access for feedback-based embedded systems," Proceedings of the
SPIE - The International Society for Optical Engineering Quality of Service over Next-Generation Internet, 30 July-1 Aug. 2002, vol. 4866, pp. 73-82,
