Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0) An OFDMA MAC Protocol Aggregating Variable Length Data in the Next IEEE 802.11ax Standard Mohand MOKTEFI, Mohand YAZID, Louiza BOUALLOUCHE-MEDJKOUNE and Wissam BENLALA LaMOS Research Unit, Faculty of Exact Sciences, University of Bejaia, 06000 Bejaia, Algeria Emails: moktefi.mohand@gmail.com, yazid.mohand@gmail.com, louiza medjkoune@yahoo.fr, wissambenlala@gmail.com Abstract—A new standard of the IEEE 802.11 standard is higher modulation rate 1024-QAM (1024-Quadrature Ampli- behind the scenes whose ratification is planned for the end tude Modulation),Down-Link/Up-Link Multi-User Multiple- of the current year (2019). It is the 802.11ax or HEW (High Input Multiple-Output (DL/UL MU MIMO) and spatial reuse. Efficiency Wireless local area network) standard dedicated to the future wireless networks. This standard promised better efficiency The implementation of these techniques poses many new and throughput in more difficult use cases (dense environments) challenges to the scientists who are working to achieve. by exploiting pre-existing physical and MAC (Medium Access Because the old Wi-Fi standards do not support these new Control) capabilities and introducing new concepts, such as: features. In this work, we are interested to efficiently manage OFDMA (Orthogonal Frequency-Division Multiple Access), 1024- the OFDMA MAC communications in the next generation of QAM (1024-Quadrature Amplitude Modulation) modulation or- der, FD (Full-Duplex) communications and spatial reuse. The IEEE 802.11ax WLANs. In fact, various methods of access new OFDMA modulation technology divides the transmission to the medium based on OFDMA have been proposed by channel into sub-carrier groups (known as: RUs for Ressources the scientific community, each aimed at optimizing the use Units) to which up to 9 users can simultaneously access a 20 of subcarriers and improving the transmission rate in a dense MHz channel. In order to improve the user data rate in dense environment. areas, an OFDMA-based MAC access method is proposed in this paper by enabling the principle of aggregating frames of The remainder of this paper is outlined as follows. Section variable lengths while ensuring synchronization. Our proposal is II introduces the OFDMA and OFDM technologies. Section followed by simulation results to demonstrate the improvement III introduces the main research works about multi-user MAC in throughput that it offers. protocols based on OFDMA technology. In Section IV, we Keywords—IEEE 802.11ax HEW, OFDMA, Medium Access, describe our OFDMA MAC protocol. Simulation results are Variable Length Data, Aggregation, Simulation and validation. given in Section VI, followed by Section VII which concludes this paper. I. I NTRODUCTION Wireless technology has become widespread on virtually II. BACKGROUND all user devices, as well as any inhabited deployment (homes, parks, airports, stadiums, etc.) since its arrival on the industrial The main change in the 802.11ax standard is the intro- market. However, users who are increasingly demanding, the duction of OFDMA technology in both downlink and uplink number of connections and bandwidth intensive applications transmissions. OFDMA makes it possible to multiplex more are growing. This increase will threaten the technology in its users in the same bandwidth. This is possible by allocating a future growth to no longer serve these customers effectively. contiguous subset of the sub-carriers (minimum 26, maximum In addition to increased reliability, future networks will need 996) of the available spectrum for each user. This means to offer greater wireless capacity. This is where the sixth that the existing 802.11ax channels (20, 40, 80 and 160 generation of Wi-Fi (Wireless-Fidelity) comes in. It’s the IEEE MHz wide) are divided into narrower subchannels with a pre- 802.11ax standard. defined number of subcarriers. The allocated amount is referred to as the Resource Unit (RU) and it is allocated to users The new 802.11ax standard, also known as HEW (High based on channel conditions and service requirements. The Efficiency Wireless local area network), has the ambitious Figure 1 illustrates the allocation of RUs by the AP (Access goal of quadrupling average throughput per user in dense Point). By using OFDM the entire channel is allocated to a areas; it is an evolutionary improvement of the 802.11ac single user, however using OFDMA several users can transmit standard. The standard has submitted three preliminary drafts simultaneously. The basic principle of OFDM and OFDMA is 802.11ax since its launch in May 2014, namely D1.0, D2.0 illustrated by Figure 2. and D3.0 and provides for finalization by the end of the current year (2019) [1]. While the new 802.11ax standard 26 26 26 26 26 26 26 26 26 is designed to maximize network efficiency, it also provides a better experience for traditional wireless LANs and more 52 52 26 52 52 predictable performance for advanced applications, such as: 4K 106 26 106 video, Ultra HD, wireless office, Internet of Things (IoT), etc. 242 tones IEEE 802.11ax will achieve its goals taking into account the following key features: Orthogonal Frequency-Division Multi- Fig. 1. Configuration of RUs on a 20 MHz band [2]. ple Access (OFDMA), adoption of Full-Duplex transmissions, current OFDMA) for the new generation WLAN. The C- OFDMA method provides improvements to the H-OFDMA method of the authors [4], [5] in terms of throughput. In fact, the H-OFDMA method for a transmission of k data results in a total of k acknowledgments transmitted sequentially to the k stations, plus an overload in transmission planning. For that, the authors to [6] have opted for a method allowing simultaneous transmissions in order to reduce this overload in data and in acknowledgments. The C-OFDMA method takes place in three phases: the Sub-channel Request (SR) phase, the Sub-channel Assignment (SA) phase and the Data- Transmission (DT) phase. T. Uwai et al. [7] highlighted on an adaptive Backoff Fig. 2. Configuration of RUs on a 20 MHz band mechanism for OFDMA random access with a finite service period in IEEE 802.11ax. Based on OFDMA random access research in the IEEE 802.11ax standard, backoff parameters In downlink transmissions, an AP can increase the power should adjust to network conditions. In 802.11ax, the service of some RUs while allowing weak users to maximize downlink period (SP) defines the operating time of the OFDMA random bit rates in the Basic Service Set (BSS), by diverting power access and the access point is supposed to initiate random from powerful user. On the other hand, the uplink OFDMA access. When SP ends, the stations should wait for the next gains are mainly due to the aggregation of multiple users. Each SP to transmit their packages. In dense environments, this leads user transmitting on his assigned RU, which contributes to a to a discontinuity, including a probability of packet loss due higher signal (SNR, signal-to-noise ratio) at the level of the to high latency, and a degradation in performance. For this, AP. In general, STAs (stations) have lower output power output the authors of [7] introduced an adaptive Backoff algorithm than APs, and this power asymmetry reduces uplink throughput for OFDMA random access using an analytical performance and may also limit the BSS range. Uplink OFDMA can be used model. They show that maximum throughput can be achieved to compensate for such power asymmetry [3]. even in dense environments by adjusting the parameters of the Backoff. The experiment is carried out on a UL-MU transmission protocol based on the OFDMA random access III. R ELATED W ORK and divided into three phases: the transmission request phase Recently, several OFDMA MAC protocols for the next (TR), the UL-MU frame transmission phase (UL-MU frame generation Wi-Fi have been proposed. A summary of some Transmission). research on the IEEE 802.11ax future standard, based on J. Lee et al. [8] suggested a new hybrid MAC protocol (H- OFDMA technology, is presented in this section. The research MAC) designed to increase channel utilization in OFDMA. studies focused above using different concepts and principles It is a protocol that relies mainly on a centralized approach to design OFDMA-based MAC access methods by adopting that also allows random access in a game. Access to the the different centralized and/or randomized scheduled access medium is controlled by three types of messages defined by controls for multi-user downlink/uplink transmissions. these authors [8] and which are: (1) Request-to-Multiple-DL P. Nasiopoulos et al. [4], [5] have proposed a hybrid (RMD), this message is sent on the channel to identify each MAC protocol based on OFDMA and CSMA/CA (H-OFDMA, candidate station for DL transmission,(2) Clear-to-Receive- Hybrid OFDMA) to increase the throughput of the next with-UL-Request (CRU), when receiving the RMD message, WLAN generation. H-OFDMA uses two frame transmission the stations designated in the DL transmissions can respond by phases: the Transmission Opportunity Request(TR) phase and sending a CRU message in their corresponding sub-channels, the Scheduled Transmission (ST) phase. The RTS packet is they may also carry information on UL requests for transmis- sent randomly by applying a CSMA/CA scheme in the TR sions, (3) Request-to-Multiple-UL (RMU), the RMU control phase. The adoption of OFDMA in the H-MAC method can message is only a trigger for UL transmissions sent by the cause a conflict between the users who are candidates for access point to the stations having requests for transmissions in a transmission when the number of sub-channels is smaller UL. As a result, the H-MAC protocol allows a random access than the number of users. Thus, several stations compete for portion for stations not designated by the RMD message in each subchannel which produces a probability of collision. the network and the RMD message is sent over the entire As a result, H-OFDMA prompts the use of CSMA/CA in transmission channel, stations not designated by the latter can the TR phase to send the RTS packet to solve the collision access randomly to one of the subchannels of the transmission problem. The access point then sequentially schedules the data channel by decoding the available free sub-channel addresses transmissions of the stations having sent their RTS packets defined in RMD and transmit their UL requests by sending the correctly and responds to them by sending the CTS packet CRU message. in the broadcast. The stations transmit their data sequentially according to the planning of the AP and each station occupies The operation of the H-MAC protocol is summarized as all the sub-channels during the transmission of its data. follows (with interframe expectations between transmissions): the access point sends the RMD message containing the G. Haile et al. [6] proposed a competing MAC protocol requests for DL transmissions to the designated stations over based on OFDMA and CSMA/CA named C-OFDMA (Con- the entire transmission channel, the stations designated by Largest frame HIFS RMD and the stations having requests UL (random access) SIFS SIFS SIFS SIFS SIFS SIFS Sub- send the CRU message to the access point in a first phase channel CRU STA A STA C - > UL DTA named Control Period, the access point transmits its data CRU STA B DL MPDU ACK -> STA A STA D -> packets to the stations designated by RMD and who responded CRU STA C RMU UL DATA by CRU and the stations acknowledge the good reception of RMD STA C 20 Mhz CRU STA D STA E -> DL data by sending ACKs each, during a phase qualified in STA A STA D UL DATA DL A-MPDU BLOCK H-MAC by Downlink Data Transmission Period. The access STA B -> STA D BA STA E ACK point then loads the UL requests through CRU, stacks them STA C CRU STA E STA F STA F -> UL DATA in the RMU message and transmits them to the designated CRU STA F DL M-PDU ->STA C STA G stations, all over the channel during the trigger phase (Trigger BA STA G -> Period) so that finally the stations having UL requests trigger CRU STA G UL DATA the transmission of their UL data and the AP responds with a BA over the entire channel to acknowledge the UL data in the Random Access STAs’ CRU Time last phase Uplink Data Transmission Period. Control Period DL aggregated-Data Transmission Period Trigger Period Uplink Data Transmission Period Padding bits The following table summarizes the main features of the studied protocols: Fig. 3. Operation steps of the A-VLD-MAC protocol. TABLE I. C HARACTERISTICS OF THE STUDIED PROTOCOLS . Articles CTS/RTS DL UL MU-OFDMA Centralised Distributed to the different stations are all of the same length [9] X X X X [4], [5] X X X X induces us to H-MAC. Thus no aggregation can be [6] X X X X X done. [7] X X X X [10] X X X X • Case 2: ∃i 6= M AX, M AX − LON G1i > 0 for [8] X X X X X X all stations there is at least one packet of length less than M AX. The occurrence of this case allows us to introduce the aggregation mechanism. To do this, the IV. AGGREGATED VARIABLE L ENGTH DATA FOR AP is looking at this point in the rest of the data of OFDMA COMMUNICATIONS each queue, the longest packet can be inserted into It is found that the adoption of H-MAC under variable X, where X is the difference between M AX and length data is inefficient at all, given the time lost to transmit LON G1i (see the equation eqref eq1), within the stuffing bits instead of raw data. However, an improvement limit of not exceeding the transmission time of the in the efficiency of the H-MAC method is demonstrated longest frame M AX (see the equation (2)), where by adopting our proposal called Aggregated-Variable Length tsM AX is the transmission time of the data M AX, DATA-MAC (A-VLD-MAC). The illustration of the A-VLD- tsLON G1i is the transmission time of the first packet MAC method is detailed in the next subsection. LON G1i and tsLON G2i is the transmission time of the second packet LON G2i that could be inserted into A. A-VLD-MAC X for the station i. The proposed A-VLD-MAC multiple access method makes M AX − LON G1i = X. (1) changes in the DL data transmission phase, with the aim of tsM AX ≥ tsLON G1i + tsLON G2i . (2) improving the transmission rate by exploiting the OFDMA RUs effectively. The improvement in fact consists of intro- • The AP encapsulates the MPDU packets, aggregated ducing the aggregation mechanism of the data frames to the A-MPDU packets if they exist, and forwards the H-MAC method, by considering the variable length of data. packets to the specified stations. The following diagram shows the different frames sent in A- VLD-MAC. • The stations having received the frames sent by the AP, answer by ACK or BA for the MPDU and A- The sending of MPDU and A-MPDU frames is done by MPDU respectively. management in the access point. This management is detailed as follows: The remainder of the H-MAC method is unchanged for the remaining period (control period, trigger period, and UL data • After receiving the CRU frames, the access point loads transmission period). The different conditions for the sequence the first data to be sent in each queue for each station of the phases are exposed in the flowchart given in the Figure and searches for the data having the maximum length. 4. • The AP verifies if the data to be transmitted for the different stations are not all of the same length, one V. S IMULATION R ESULTS AND A NALYSIS of the following two cases will be applied: We have used the C programming language under Linux • Case 1: ∀i 6= M AX, M AX − LON G1i = 0 where operating system for implementing and assessing the perfor- M AX is the maximum length and LON G1i is the mance of the proposed A-VLD-MAC method designed for length of the packet to send to the station i. Falling in optimizing the OFDMA MAC communications under variable the case where the different packets to be transmitted length data. The choice of programming language is made in packets in both methods, a simulation is implemented. The Wit HIFS and start A-VLD-MAC results of this simulation are given in Figure 5. According H- MAC A-VLD-MAC Has data Tx RMD to send? for UL 500 450 Tx RMD 400 with Rx CRU? DL STA info. 350 300 Mbits/S Rx Tx MRU Rx UL CRU? with UL data & 250 STA info. Tx ACK 200 DL 150 STA’s CRU? 100 50 LONGi = MAX ? 0 1 2 3 4 5 6 7 8 9 Stations Tx DL DATA & CRU with LONG2i < X Rx ACK UL Info? Fig. 5. Average throughput. TsMAX ≥ tsLONG1i+ tsLONG2i ? to the results of the simulation, the A-VLD-MAC method considerably improves the average throughput compared to the H-MAC method. The average bit rate can reach 453 Mbps in Tx MPDU +A-MPDU & A-VLD-MAC compared to 319 Mbps in H-MAC, a difference Rx ACK + BA Resp of 134 Mbps. These results show the effectiveness of the A- VLD-MAC method in dense networks. Fig. 4. Flowchart of the transitions of A-VLD-MAC. B. Average throughput per user Responding to user requirements while providing better relation to the simplicity, flexibility and speed of the language. service in high density areas is one of the goals of the new We have used a simulation environment with a 20 MHz trans- IEEE 802.11ax standard, and providing better throughput is mission channel, supporting 9 simultaneous users in OFDMA one of the important services. The Figure 6 represents the (RU = 26 subcarriers). The PHY and MAC parameters that results of the simulation of the average flows per user in each we have put forward to evaluate the performance are defined of the two H-MAC and A-VLD-MAC methods. in the Table II. TABLE II. IEEE 802.11 AX PHY AND MAC PARAMETERS . H-MAC A-VLD-MAC 70 Parameters Signification Value Channel bandwidth bandwidth width 20 MHz 60 HIFS inter-frame time H-MAC 25 s SIFS inter-frame time 16 s 50 PHY header PHY header transmission time 36 s MAC header length of the MAC header 320 Bits 40 Mbits/S Data rate Data rate transmission 65 Mbps Basic rate Data rate of the overhead 6 Mbps 30 Del Pad Delimiter size and padding of A-MPDU packets 56 Bits ACK ACK size 112 Bits 20 BA size of ACK block 320 Bits 10 The performance metrics that we have computed to eval- 0 1 2 3 4 5 6 7 8 9 uate the performance of our A-VLD-MAC proposal are: the Stations average throughput, the average throughput per user, and the bit loss rate of the DL transmissions. Given the proposed Fig. 6. Average throughput per user. method A-VLD-MAC is an improvement of the H-MAC method with variable packet lengths, the simulation results obtained for the A-VLD-MAC method are compared to those We can see clearly in the figure 6 above, that the average of the H-MAC. throughput per user in the A-VLD-MAC method is consider- ably improved compared to the results of the H-MAC method. A. Average throughput Apart from the average throughput per user offered for a single station that is equal in both methods (specific case A-VLD- The average rate determines the speed of data transmission. MAC), the average throughput per maximum user achieved and In order to evaluate the rate of DL transmissions in the network which is almost stable by increasing the number of stations in by varying the number of stations and the lengths of the A-VLD-MAC is 56 Mbps against 50 Mbps for H-MAC and can be downgraded to 33 Mbps by increasing the number of R EFERENCES stations. [1] E. Khorov, A. Kiryanov, A. Lyakhov, and G. Bianchi, “A tutorial on ieee 802.11ax high efficiency wlans,” IEEE Communications Surveys Tutorials, vol. 21, no. 1, pp. 197–216, Firstquarter 2019. C. Bit loss rate [2] K. Wang and K. Psounis, “Scheduling and resource allocation in A large amount of bits are sent using the H-MAC method 802.11ax,” in IEEE INFOCOM 2018 - IEEE Conference on Computer Communications, April 2018, pp. 279–287. with lengths of varying sizes to synchronize transmissions. [3] D. Deng, Y. Lin, X. Yang, J. Zhu, Y. Li, J. Luo, and K. Chen, “Ieee The amount of stuffing bits sent in a DL transmission varies 802.11ax: Highly efficient wlans for intelligent information infrastruc- according to the lengths of the transmitted data for each ture,” IEEE Communications Magazine, vol. 55, no. 12, pp. 52–59, Dec user. The charge in stuffing bits is reduced by adopting our 2017. proposition A-VLD-MAC. The following Figure 7 shows the [4] H. Alnuweiri, Y. P. Fallah, P. Nasiopoulos, and S. Khan, difference between the rate of loss in stuffing bits using the “Ofdma-based medium access control for next-generation wlans,” H-MAC method and the method A-VLD-MAC for the same EURASIP Journal on Wireless Communications and Networking, vol. 2009, no. 1, p. 512865, Feb 2009. [Online]. Available: lengths of data sent to the same number of stations in both https://doi.org/10.1155/2009/512865 methods. [5] Y. P. Fallah, S. Khan, P. Nasiopoulos, and H. Alnuweiri, “Hybrid ofdma/csma based medium access control for next-generation wireless H- MAC A-VLD-MAC lans,” in 2008 IEEE International Conference on Communications, May 60 2008, pp. 2762–2768. [6] G. Haile and J. Lim, “C-ofdma: Improved throughput for next gener- 50 ation wlan systems based on ofdma and csma/ca,” in 2013 4th Inter- national Conference on Intelligent Systems, Modelling and Simulation, 40 Jan 2013, pp. 497–502. Taux de perte % [7] T. Uwai, T. Miyamoto, Y. Nagao, L. Lanante, M. Kurosaki, and H. Ochi, 30 “Adaptive backoff mechanism for ofdma random access with finite service period in ieee802.11ax,” in 2016 IEEE Conference on Standards 20 for Communications and Networking (CSCN), Oct 2016, pp. 1–6. [8] J. Lee and C. Kim, “An efficient multiple access coordination scheme 10 for ofdma wlan,” IEEE Communications Letters, vol. 21, no. 3, pp. 596–599, March 2017. 0 [9] Q. Qu, B. Li, M. Yang, and Z. Yan, “An ofdma based concurrent 2 3 4 5 6 7 8 9 multiuser mac for upcoming ieee 802.11ax,” in 2015 IEEE Wireless Stations Communications and Networking Conference Workshops (WCNCW), March 2015, pp. 136–141. [10] G. Naik, S. Bhattarai, and J. Park, “Performance analysis of uplink Fig. 7. Bit loss rate multi-user ofdma in ieee 802.11ax,” in 2018 IEEE International Con- ference on Communications (ICC), May 2018, pp. 1–6. We can see in Figure 7 that the loss rate varies according to the length of the packets and the number of stations in the H-MAC method, it can reach a rate of '44% bit loss stuffing. On the other hand, the maximum loss rate in stuffing bits in A- VLD-MAC does not exceed ' 13 %, so a gain of '31%(44%- 13%=31% ) in data is offered using A-VLD-MAC. We also note, the stability of the loss rates in A-VLD-MAC, despite the variation in number of stations and lengths of the data, which guarantees the efficiency of A-VLD-MAC in the reduction of loss rates in stuffing bits. VI. C ONCLUSION The efficient use of transmission channel in the IEEE 802.11ax standard allows for increased throughput. The pur- pose of the proposed A-VLD-MAC method is to optimize the use of OFDMA transmissions by improving the H-MAC method while introducing the data frame aggregation mech- anism defined in the IEEE 802.11e standard. In this paper, we have detailed the operation steps of the proposed method A-VLD-MAC which is an improvement of the H-MAC [8] method, while underlining the differences between the two methods. A simulation part is also implemented in this work to demonstrate the effectiveness of the proposed solution. Obviously, the results given by the simulation of the two methods assert that the proposed method A-VLD-MAC is more efficient than H-MAC by increasing the transmission rate and reducing the rate of loss in stuffing bits.